Context Emergence through Stimulus Competition in Extinction Learning

Context Emergence through Stimulus Competition in Extinction Learning | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Context Emergence through Stimulus Competition in Extinction Learning Juan Peschken This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6146552/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract In extinction learning, contextual renewal occurs when an extinguished response reemerges due to a contextual shift. But what determines which stimuli function as context? Rather than being an inherent property of the environment, we propose that context emerges from competition among multiple stimuli. Using a novel paradigm in pigeons, we systematically assessed the relative influence of local, spatial, and global stimuli in driving contextual renewal. Our results reveal that renewal depends on the competitive dynamics between these stimuli, rather than any single cue acting as context by default. This challenges traditional definitions of context as a passive background and suggests that context should be operationalized based on its interaction with other available cues. Understanding these competitive mechanisms provides a framework for studying how attention and learning shape contextual control in extinction. Psychology Extinction learning ABA renewal Context Pigeons Operant conditioning Stimuli competition Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction In the process of adapting to the environment, organisms are continuously exposed to a multitude of stimuli that compete for attention and relevance. Some of these stimuli are selected and prioritized, leading to learning processes in which certain stimuli are sought out for their beneficial outcomes, while others are avoided to prevent harmful consequences. But how does an organism determine which stimuli are useful and contain informative value? This selection mechanism has been studied extensively, leading to the development of numerous theories on learning and attention (1–4). Today, we know that each organism may have a particular approach to determining this competition. However, in general, the selection of relevant stimuli depends on two main aspects: internal factors associated with the organism, such as sensory capabilities and the significance of each stimulus according to its biological needs, and external factors, related to the stimulus itself, including the relative salience compared to other stimuli (5), its duration and stability (continuity), (6) the temporal proximity to associated consequences (contiguity) (7), and the reliability of its occurrence with those consequences (contingency) (8). If these elements are relevant for learning the relationships between stimuli, behavior and outcomes, they should also play a crucial role when those relationships cease to function and an ‘update’ is necessary. Central to this capacity is extinction learning, a process through which previously learned associations are gradually diminished in response to the omission of expected outcomes. However, this adjustment is not always straightforward. While extinction learning is crucial for the suppression of maladaptive behaviors, such as pathological fear responses or problematic conducts (9–11), it does not necessarily result in a permanent loss of learned associations. Extinction learning is best understood as a new learning process that conditionally inhibits previous associations. Several of these conditions are evident in the phenomenon of renewal (12, 13), where extinguished behaviors reemerge when an individual changes the context that no longer supports the learned association. The renewal effect has been studied in various experimental paradigms, most commonly the ABA design (14), where animals first learn an association in context A, undergo extinction in context B, and are then tested in the original context A. Renewal is observed when the conditioned behavior reappears in response to the return to context A. While this classic framework has been widely used, the precise nature of context—what stimuli are involved, how they interact, and how their influence on behavior is determined—has traditionally been neglected. Context has often been defined as a passive background of environmental cues, such as spatial or sensory features that surround a learning experience (15–18). However, a growing body of research suggests that context is not merely a fixed environmental backdrop. Instead, context appears to be learned dynamically from stimuli present when the original association ceases to function (19–21). Critically, when learning occurs among multiple individual stimuli, competition arises between them. This competitive view of context implies that no single stimulus inherently serves as the context, nor is context simply a compound of all available stimuli. Rather, it emerges from the relative influence of competing stimuli, each contending for control over the learned behavior. To further explore this framework, we designed a novel experimental approach in pigeons aimed at systematically assessing how different types of stimuli—spatial, local, and global—compete to serve as the context for extinction and renewal (22). During extinction in context B, all three types of stimuli were altered, but only one returned to its original form in the test context (A’) forming an ABA' paradigm. All stimuli were controlled to have the same level of contingency and contiguity, while specific parameters of continuity and saliency allowed for meaningful comparisons with previous reports. By manipulating these conditions, we aimed to determine which cues are more likely to be selected as the context and how their competitive dynamics influence the renewal of conditioned responses. Materials and methods Subjects. We conducted the experiment on seven pigeons ( Columba livia ), consisting of four females and three males of unknown age, sourced from local breeders and individually housed in a colony room. Throughout the experimental procedures, the birds had unrestricted access to water and were maintained on a controlled feeding protocol. They received food rewards during the experimental sessions. These pigeons are part of a group of animals that provided a complete dataset, as detailed in this study pre-registration (22). For this experiment, three animals were completely naïve, while four had previously participated in a similar protocol. There was at least a three-month gap between the two protocols. All experimental procedures complied with the German guidelines for the care and use of animals in science, as well as the European Communities Council Directive 86/609/EEC on the care and use of animals for experimental purposes. The study was approved by the ethics committee of the State of North Rhine Westphalia, Germany. Experimental setup . The experimental setup and initial training procedures were conducted as previously described in (21) . Briefly, we used a plus-shaped arena (3.5m x 3.5m) composed of four identical arms, each illuminated by programmable LED lamps. At the end of each arm, a 43” monitor displayed geometrical shapes as landmarks. Overall, the arena contained eight distinctive locations where the birds could interact and perform the task. A camera was installed at the center of the arena’s ceiling to observe and record the birds’ behavior. A water bowl was always present in the center of the arena at floor level. The layout provided eight distinct touchscreen locations for the birds to interact with and receive rewards, with one active location at a time. The open design allowed the pigeons to move freely, including flying in and out, mirroring their natural foraging behavior. Three contextual dimensions in the arena setup. The configuration of the arena employed in this experiment allowed controlling context with three distinct stimuli (Fig. 1A).The spatial context , defined by the physical position of the animal in the arena, enriched by a large landmark displayed in the 43” monitor at the end of each arm, a geometrical figure (triangle, hexagon, horizontal bar, quarter circle), the landmarks were always displayed in the same monitors, remaining stable throughout the experiment. The local context, which was the background color of the 7” active touchscreen on which cue-stimuli were displayed and selected. The colors followed the RGB color model, overall, six different colors were used, the main three colors and their most extreme combinations: Red (1, 0, 0), Green (0, 1, 0), Blue (0, 0, 1), Yellow (1, 1, 0), Cyan (0, 1, 1) and Magenta (1, 0, 1). The global context , designated by the background color of the 43” monitor at the end of the arm where the task was being performed, four colors were used, Red (1, 0, 0), Green (0, 1, 0), Blue (0, 0, 1) and Yellow (1, 1, 0). The birds were randomly assigned to face a specific context in their first session and then continued alternating contexts in each subsequent session, each context was tested four times. Overall, the seven pigeons underwent a total of 84 sessions (12 each). The capacity of each stimulus to function as a context was evaluated by counting the number of renewal responses when the respective context returned to the acquisition array after extinction. Custom MATLAB code (Mathworks, Natick, MA, USA) was used to control stimulus presentation and contextual features, utilizing the open toolbox for behavioral research (23) and the Psychophysics toolbox (24). Behavioral protocol. The pigeons performed a forced choice discrimination task, originally adapted from (25–27) and similar to the employed in (21) (Fig. 1B). In each session, we used two pairs of choice stimuli (S+ and S-), one familiar and one novel. The familiar pair was pre-trained and remained constant across all sessions, serving as a control and maintaining task engagement throughout the different phases of a session. Conditioned responses to the familiar stimuli (S+) were always rewarded, enabling us to control satiety levels and potential fatigue during the session. The novel pair, unique to each session and never presented before, were the critical stimuli for testing acquisition, extinction, and renewal. Both familiar and novel stimuli, as well as their screen locations, were presented in a pseudo-randomized and balanced order. The session began with the pigeon flying freely into the arena, searching for the active touchscreen to initiate the task. Each trial began with the presentation of an initiation stimulus, a black dot centered on a touchscreen, displayed for a maximum of ten seconds. The trial was initiated by a single peck to this stimulus, followed by a one-second delay before the choice stimuli appeared on the screen for up to six seconds. Three possible interactions could occur: a peck on the S+ (‘correct response’) resulted in a reward of one food pellet and the illumination of the LED below the feeder for 0.5 seconds; a peck on the S- (‘alternative response’) resulted in a black screen for one second, no food reward, and an additional one-second delay before the next trial began; no interaction with either stimulus during the presentation period (‘no response’) resulted in no food reward or additional feedback. Consecutive trials were separated by an inter-trial interval (ITI) of four seconds. Once all trials at a location were completed, a black-and-white checkerboard appeared, signaling the animal that the current touchscreen was inactive, encouraging it to search for another site Three distinct phases, acquisition (A), extinction (B), and renewal (A) took place within each session, all with a specific contextual array including all three contextual stimuli (Fig. 1C), effectively establishing an ABA protocol (28, 29). During the acquisition phase, subjects learned the S+/S− association for the novel stimuli through trial and error. The physical array of the arena was constant during each acquisition session (A). For the local context, all touchscreen backgrounds were white, while for the global context, all monitor backgrounds were black. For the spatial context, to ensure balanced exposure across sessions, we predetermined a random order of the eight possible locations. This order was followed for all subsequent sessions, ensuring that after completing the full cycle of eight locations, the repeated spatial context was temporally distant from the previous first exposure. Acquisition consisted of a minimum of 80 trials (40 for each pair of stimuli) and continued until the subject met two criteria: completing the minimum trial count and reaching 85% correct responses in the last 20 trials for both familiar and novel stimuli. After the acquisition phase was completed, a one-minute ITI was implemented. The bird then moved freely, looking for the next active touchscreen to continue the task in the extinction phase. The extinction phase followed in a new contextual configuration, altering all three context stimuli. Animals performed the task on a different touchscreen located in a separate arm of the arena (spatial context), with the touchscreen background changing to a new color (local context). This change was triggered only after the trial was successfully initiated, the birds began the trial with the known white background, and the color change occurred only after initiation, in conjunction with the appearance of the choice stimuli (S+ and S-).The monitor background in the new arm also changed to a different color (global context), this change was stable thought the extinction phase. During this phase, responses to the novel stimuli no longer elicited feedback (neither food nor timeout); instead, interactions with S+ and S− simply cleared the screen, returning it to the previous white background and initiated the standard ITI. The extinction phase consisted of a minimum of 80 trials (40 familiar, 40 novel) and concluded only when subjects met two criteria: completing the minimum trial count and achieving 85% correct responses on familiar stimuli and 85% extinction on novel stimuli, both calculated over the last 20 trials. Extinction was defined as trials where the bird either responded to the S− or omitted a response entirely, while responses to S+ were classified as a failure to extinguish. Once the extinction phase was complete, a second one-minute ITI followed. Here, the animal roamed through the arena in search of the final active screen. In the renewal phase, only one of the three contextual stimuli reverted to its original acquisition configuration, while the remaining stimuli stayed as they were during extinction (A’). This setup created a ‘competition’ among the contextual stimuli, enabling a direct comparison of renewal effects across different context types. During renewal, responses to the novel stimuli continued without feedback, mirroring extinction, to observe the renewal effect. Renewal was induced one minute later, controlling for potential recovery effects like reinstatement and spontaneous recovery, ensuring the observed responses were due to the renewal effect. The renewal phase comprised a minimum of 40 trials (20 familiar, 20 novel) and concluded only when the subject met two criteria: completing the minimum trial count and achieving 85% extinction responses in the last 20 trials, defined as the absence of responses to the unrewarded S+. Statistical analysis. The sample size for this study was determined through a priori power analysis using G*Power 3.1 software (24). Based on a preliminary pilot study, we assumed an effect size of dz = 1.09, a significance level of α = 0.05, and a power of 0.8. Data analysis was performed using custom MATLAB code (25), while GLMM-specific analyses were conducted in R Statistical Software (26) with the packages lme4 (27), flexplot (28), and DHARMa (29). To assess successful acquisition, extinction, and renewal, we recorded the number of responses and compared them to pre-defined criteria: 17 out of 20 correct responses (i.e., 85%) in the acquisition phase, and 17 out of 20 extinction responses (either no response or alternative responses) in the extinction and renewal phases. This criterion was considered statistically significant (p < 0.05) under a cumulative probability within the binomial distribution. The probability of obtaining at least 17 correct responses out of 20 trials by chance, assuming a 50% random choice, was calculated to be 0.00020123. We used multiple generalized linear mixed models (GLMMs) to analyze the response patterns to different types of stimuli across experimental phases and to evaluate the strength of the renewal effect for each contextual stimulus. In total, three distinct models were fitted. The first model examined the change of familiar and novel stimuli across the different phases. For this purpose, a binary outcome variable (Response: correct = 1, incorrect = 0) was modeled based on three main fixed effects and their three-way interaction. These fixed factors included experimental phase (Phase: acquisition, extinction, and renewal), target stimulus (Stimulus: familiar and novel), and trial within phase (Trial: trial count for each Stimulus within each Phase). The random-effects structure of the model included Sessions nested within Subjects. To account for the variable session lengths in our design, the specific interaction between Stimulus and Trial was incorporated as a random term. The model was fitted using a Binomial distribution and the logit link-function. M1 = Response ~ Stimuli*Phase*Trial + (Stimuli:Trial | Subject:Session) In the second model, the number of renewal responses to the novel stimuli was analyzed based on two main fixed factors. These factors were the contextual stimulus (Context: Global, Spatial, and Local) and the contextual session number, representing each test within a specific context (CtxtSess: 1, 2, 3, and 4). The random-effects structure included Sessions nested within Subjects. The model was fitted using a Poisson distribution and the log link-function. M2 = Renewal ~ Context + CtxtSession + (1 | Subject:Session) The third and final model aimed to analyze the temporal dynamics of renewal onset, considering only sessions where renewal responses were observed. Sessions that did not produce renewal were excluded from this analysis. For this purpose, we established a latency metric, defined as the number of trials required to initiate renewal responses. This was then analyzed as a function of two main factors: Context (Global, Spatial, and Local) and the amount of renewal in each session. To better capture the influence of each context on latency, Context was included as a random factor. The random-effects structure likewise accounted for sessions nested within subjects. Since the model dealt with count data, it was fitted using a Poisson distribution with a log link function. M3 = Latency ~ Context + Renewal + (Context | Subject:Session) In addition to the analysis performed on the data acquired in this experiment, two additional contrasts were made with previously published data (21). These comparisons, (refer to supplementary material SI1 & SI2) examined the influence of an increase in continuity for the local context (C1) and an increase in saliency for the spatial context (C2) in the amount of renewal responses. Each context was tested four times per animal, with the same seven animals participating across studies, resulting in a total of 56 sessions in each comparison—28 for each context. The analysis described here is novel and represents the first time these specific comparisons have been conducted. In C1, the effect of increased continuity in the local context (local+) was tested by extending the background color change on the touchscreen, which was not only present during the choice stimulus (as implemented in this study) but also during the presentation of the initiation stimulus. On the other hand, C2 assessed the effect of enhanced saliency for the spatial context (spatial+). In this comparison, in addition to the different locations enabled by the arena, spatial information was enhanced by integrating the global context, using monitors with a constant color that corresponded to the illumination in each arm of the arena, making the different arms more recognizable. Both comparisons employed the same model structure where renewal responses were fitted as a function of two main factors, the contextual stimulus (Context: Local and Local+ for C1, or Spatial and Spatial+ for C2) and contextual session number, counting each test within a specific context (CtxtSess: 1, 2, 3, and 4). The random-effects structure included Subjects to accommodate for the repeated measurements design. Both models were fitted using a Poisson distribution and the log link-function. C = Renewal ~ Context + CtxtSession + (1 | Subject) Results We observed the behavioral responses of seven pigeons in a two-stimulus discrimination task using a within-session acquisition-extinction-renewal paradigm (ABA) in an open arena. Three distinct types of contextual stimuli were implemented: spatial context, local context (defined by the background color of the active touchscreen), and global context (designated by the background color of the large monitor at the end of the active arm in the arena). To investigate the contextual control of individual stimuli, only one of the three contextual stimuli reverted to its acquisition state during the renewal test (ABA’). Model 1 explored the dynamic of conditioned response of familiar and novel stimuli across the three distinctive phases (Fig. 2 ). Familiar stimuli had a high initial probability of CR in ACQ, \(\:\widehat{P}\) = 0.96 (β = 3.415, t= 15.401, p < 0.001) and remained unaffected across the change of phases, for EXT (β = 0.482, t= 1.618, p = 0.105) and for REN (β = -0.007, t= -0.026, p = 0.979). Trial had a marginal positive effect for familiar only in ACQ, \(\:{\Delta\:}\widehat{P}\) = + 0.0008, (β = 0.026, t= 2.719, p = 0.006), and no significant deviations in EXT and REN. In contrast, the novel stimuli had a lower initial probability of CR in ACQ, \(\:\widehat{P}\) = 0.63, (β = -2.863, t= -12.503, p < 0.001) but a stronger positive effect of trial representing the expected learning curve, \(\:{\Delta\:}\widehat{P}\) = + 0.034, (β = 0.126, t= 8.389, p < 0.001). As anticipated, trial had a negative effect for novel in EXT, \(\:{\Delta\:}\widehat{P}\) = − 0.038, (β = -0.319, t= -17.814, p < 0.001) and REN, \(\:{\Delta\:}\widehat{P}\) = − 0.030, (β = -0.277, t= -12.06, p < 0.001), demonstrating the decay of performance due to violation of reward expectancy both in extinction and renewal. Model 2 examined the influence of individual contextual stimuli on the return of conditioned responses, reflecting the strength of the renewal effect (Fig. 3 ). All three contextual stimuli elicited renewal, albeit to different degrees. Without the effect of session, the global context yielded 2.7 responses as a pure effect (β = 1.021, t = 3.427, p < 0.001), the spatial context generated 13.9 responses (β = 1.614, t = 5.447, p < 0.001), and, unexpectedly, the local context produced the highest response count of 29.2 (β = 2.354, t = 8.19, p < 0.001). There was a significant decay in renewal for all contexts across repeated tests, with an average reduction of 48.5% of responses per session (β = -0.663, t = -7.491, p < 0.001). When analyzing the temporal dynamics of renewal, Model 3 identified differences in the typical latency across contexts (Fig. 4 A). Animals exhibited the shortest latency in the local condition, specifically when the local context matched the one from the acquisition phase, with an average renewal onset at trial 1.89 (β = 0.638, t = 2.590, p = 0.012). In the spatial condition, renewal onset was consistently later, with an average latency of 3.87 trials (β = 0.717, t = 3.104, p = 0.003). The global condition showed the longest latency, with renewal beginning on average at trial 9.57 (β = 1.620, t = 5.613, p < 0.001). Interestingly, the model revealed that the renewal predictor showed no relationship between the number of renewal responses in a session and latency, indicating that the "intensity" of renewal for each context was independent of when responses began (β < 0.001, t = 0.041, p = 0.967). This finding sparked a new analysis. Intuitively, one might expect that if an animal is ‘highly certain’ that the context has changed and that the reinforcer will become available again, renewal responses should emerge quickly and be more robust. Conversely, ‘lower certainty’ about the context change should result in slower and weaker renewal responses. However, this was not the case in our data. While context modulated both the number of renewal responses and their typical onset, there was no relationship between response intensity (i.e., the duration or total number of responses) and latency. Remaining agnostic about the underlying mechanisms driving this behavior, but with the intention of gaining more insight into this effect, we applied a Kernel smoothing function to the latency data to estimate the potential distribution of latencies across contexts (Fig. 4 B). To assess whether the resulting distributions differed significantly, we first simulated 1,000 data points from a linearly spaced vector spanning the minimum and maximum latency values and interpolated them to ensure that the resulting Kernel distributions were assessed on the same data pool. We then performed pairwise comparisons using a Bonferroni-corrected two-sample Kolmogorov-Smirnov test. The corrected p-value threshold for significance was set at 0.016. All three distributions were significantly different from each other: Local to Spatial (D = 0.721, p < 0.001), Local to Global (D = 0.712, p < 0.001), and Spatial to Global (D = 0.309, p < 0.001). This indicates that not only do the contexts differ in the typical renewal onset latency, but also in the range of trials in which renewal responses are expected to occur. When comparing results across studies, we found that increasing continuity (i.e., the duration of stimulus presentation) for the local context led to a higher number of renewal responses (Fig. 5A). In the first session the local+ produced 25.2 renewal responses (Δ + 12.1 to local) (β = 0.660, t = 7.684, p < 0.001). A similar effect was observed for the increased salience in the spatial context, where spatial+ produced 13.2 renewal responses (Δ + 6.1 to spatial) (β = 0. 619, t = 5.145, p < 0.001). These results reveal a clear trend: increasing the parameters that enhance stimulus learning leads to a higher number of renewal responses. Additionally, both comparisons also confirmed the expected decay effect with repeated testing: the number of renewal responses decreased as the number of test sessions increased. Specifically, the decay rate per session was 39.6% in each session for the local context (β = -0.505, t = -12.499, p < 0.001) and 46% for the spatial (β = -0.616, t = -10.329, p < 0.001). Discussion In this study, we expand on our new experimental approach, ABA', focusing on the competition between potential contextual stimuli during extinction learning. This method enables direct testing of each context in competition with others, using the frequency of renewal responses as a proxy to assess how effectively a specific stimulus functions as a context. Three types of stimuli were employed in this experiment: global (large distal visual cues), spatial (continuous visuospatial surroundings), and local (small proximal visual cues). All three types produced renewal, though in a differential manner, both in terms of the intensity of renewal responses and the temporal dynamics of when renewal occurred. Comparisons between the current data and previous results confirmed that manipulating learning parameters, such as salience and continuity, modulates the strength of renewal responses. Contextual variables are often studied in isolation, with a single manipulated stimulus serving as the context, or all contextual stimuli returning to the acquisition state simultaneously. This simplifies the task for subjects, as they can easily detect a clear difference between the acquisition, extinction, and renewal phases. However, this can obscure the relative effectiveness of individual stimuli as context and their influence on behavior. A small, but noticeable, change in internal states or the environment can lead to renewal effects (29, 30). In contrast, our approach directly tests how a stimulus serves as the context by creating competition among the available stimuli. But how can we then explain the results we obtained? First, it is important to emphasize that our design and the open nature of the arena ensure that the pigeon samples all available information and cannot avoid or "miss" any source. The pigeon moves freely within the arena, gaining access to both spatial and global information as it transitions between phases (acquisition, extinction, and renewal) and forages for the new location where the task takes place. Pigeons' visual and attentional capacities have been extensively studied, showing that they can simultaneously process multiple points within their visual field, even employing differential neural pathways. They exhibit simultaneous discrimination of frontal objects (tectofugal pathway) and lateral, more distant information (thalamofugal pathway) (31–33), which aligns precisely with the placement of our local and global contextual cues. As a result, when providing a response, the pigeon is simultaneously exposed to all three sources of contextual information, which would suggest that the observed differences in the contextual renewal are due to either, the ‘properties’ of the pigeons or the ‘properties’ of the stimuli. Second, our results further demonstrate how the different information presented in the context was not sampled as a passive backdrop where extinction learning occurred, as would align with the classical definition of context. Instead, our pigeons showed differential responses to each stimulus, demonstrating that each one was processed and attended to in a distinct manner. This reinforces our proposal that the ‘formation’ of context in extinction learning is an active effort by the subject to disambiguate the conflict of performing the same behavior while receiving different outcomes, context is the result of an active learning process, rather than just a passive backdrop of the environment. This is consistent with previous reports highlighting pigeons' tendency to treat different elements not as a compound, but as separate parts, processing each source of information individually (34–36). It is conceivable that, given the evolutionary history of the pigeon and its perceptual apparatus, there may be an intrinsic hierarchy that explains why local, spatial, and global stimuli generate differential renewal, a so-called ‘pigeon property’. For example, in visual discrimination and search tasks, it has been shown that pigeons process information differently from primates, being more sensitive to local rather than global stimulus features (37–39). However, studies have shown that this tendency is not static and can reverse, resulting in a preference for global elements, particularly when the exposure time to the relevant stimuli is extended (40, 41). In our paradigm, spatial and global information is presented continuously throughout the entire experimental session, unlike the local information, which is presented briefly and intermittently only during specific moments in the task. Considering the differential presentation of stimuli, the local context should be the most affected by the brief presentation times. This is further supported by the comparison with greater continuity: longer presentation times lead to a stronger response. These challenges the idea that the local context is stronger simply because it is presented for a shorter duration, as, in fact, the increased exposure appears to generate a greater response. Although we cannot rule out the possibility that some intrinsic ‘pigeon property’ contributes to the pattern of results we observed, we believe there are more compelling arguments supporting the role of ‘stimulus properties’. For instance, we have shown that the effectiveness of contexts can be modulated by the properties of the stimuli. Both local and spatial stimuli performed better when we increased their continuity and salience, respectively. Likewise, this modulation does not only go in one direction, when we reduced the contiguity of the local context, we disrupted its ability to function as a context, causing it to lose the contextual competition and fail to produce renewal (21). Additionally, other accounts have highlighted how task-related factors can influence contextual processing, especially if the context proves to be useful for task solving (42–44). Although, by definition, none of our contexts directly assist in solving the discrimination between the stimuli providing food to the pigeon, we cannot deny that the proximity of the local context to the task stimuli might increase its perceived 'relevance', as it is the only context that occurs contingently with the task. This aligns with similar findings in the study of the overshadowing phenomenon. Overshadowing is a classical phenomenon in conditioning and learning, where one stimulus becomes more influential in the learning process, overshadowing the effect of another stimulus (45, 46). This occurs when two stimuli are presented together during conditioning, but one stimulus is more salient or noticeable, causing it to dominate the learning process. As a result, the less salient stimulus fails to acquire as much associative strength or influence. There have been attempts to modulate overshadowing through stimulus duration, but these efforts have yielded conflicting results (47) or have not been successful altogether (48). In contrast, other studies examining distance and proximity variables have shown that the spatial distance between stimuli can result in one stimulus becoming more salient than the other (49, 50), This has been extended to human participants as well as pigeons (51, 52) . This aligns with our findings, where it appears more likely that the observed hierarchy is not caused by temporal dynamics related to the presentation differences of the stimuli, but rather by the proximity of the different sources of information to the pigeon. If we have a hierarchy based on the ‘stimulus properties’, it makes sense to expect that the pigeons would follow this hierarchy when faced with a context shift and the potential return of the reward. This is precisely what we observe: the three distinct contexts show differentiated latency and exhibit a staircase-like behavior. They seem to first sample the local context, followed by the spatial context, and finally the global context. Only once they reach the context that has changed do they begin their renewal responses. Pigeons seem to process one stimulus before moving on to the next, focusing on one context at a time in a predetermined manner. If this were the case, the hierarchy would also explain the differential intensity of renewal. Upon returning to the renewal phase for the local context, the pigeon experiences the local change immediately; both novel and familiar trials indicate that the context has changed and that the reward should now be back. However, if it is the spatial context that changes, meaning the pigeon actively moves to the original location, the local information is again the first to be sampled before the spatial one. The pigeon would then experience trials, both novel and familiar, indicating that it remains in the same local context of extinction, that nothing has changed, accumulating evidence that the reward has not returned and permeating the spatial context with this update. Consequently, by the time the pigeon samples the global context information, it has already gone through all previous trials where both local and spatial contexts, indicate that nothing has changed and that it continues in extinction. If the global context does exhibit any renewal response, it arrives late and lasts for a short duration. The key to understanding this staircase-like dynamic may lie in how attention is allocated across these different contexts. Organisms generally disregard contextual information when it is not relevant—such as when the task is clear and provides all the required details to obtain food. However, during extinction, when the outcome is omitted, organisms redirect their focus to the context, resulting in context-dependent processing (15, 53). It has been suggested that attention could guide the process of information search with the aim of disambiguating the absence of the expected outcome in extinction learning (54, 55). Additionally, there are references showing how contextual information can hierarchically organize pigeons' behavior (56). While other studies suggest that pigeons can learn from past experiences and become primed to allocate attention in a specific way (57). Furthermore, pigeons tend to be much more exploratory and prone to attentional switching in complex choice situations than humans (58). All of this leads us to consider that differences in stimulus properties and patterns of attention allocation and processing during context formation, could generate reasonably repeatable and predictable structures when selecting which stimuli can become the context in extinction learning. This has relevant implications not only for experimental work and basic science but also for translational efforts and clinical applications. If we could take these dynamics into account when designing experiments, we could have more control over the variables we want to be attended to and identify potential confounds in our experiments. Extending this to therapeutic applications, we could design environments where therapy is conducted to promote certain stimuli, and not others, to be selected as the context for extinction. This would be immensely useful in ensuring that after extinction, there is no relapse and reappearance of undesired effects. We could select a context that is easily generalizable to other situations, facilitating the retention of learned inhibition in various contexts. Alternatively, we could consider a ‘portable’ stimulus or context that can be carried by the patient and used during more challenging stages of exposure, thereby preventing relapse during the process. Achieving this requires meticulous control of the variables described in this study and others of great importance, such as salience. Additionally, validating attentional processes with gaze tracking or neural correlates would be imperative to truly understand how the brain resolves the competition between different contexts. In conclusion, our study provides compelling evidence that contextual stimuli compete for attention and influence renewal responses in pigeons. The differential renewal observed across local, spatial, and global contexts suggests that pigeons actively process and prioritize contextual information based on stimulus properties, providing more evidence that context is not merely a backdrop for behavior and a passive element in extinction learning, but rather an active process carried out by the organism. This process can be modulated and should be understood to improve our understanding of extinction and limit the occurrence of relapse. Planned study protocol. This study was part of a planned experimental protocol that was registered with animalstudyregistry.org, on 25th January 2023. Details of the planned protocol are available on the Animal study registry platform’s website at (22) https://www.animalstudyregistry.org/10.17590/asr.0000305. The original protocol consisted of three experimental manipulations (E1 – E3). This report covers the results of experimental manipulation E1. Results of experimental manipulation E2 and E3 can be found on Peschken et al., 2025 (4). We deviated from the original protocol with regard to the statistical analysis. We did not apply t-test and ANOVA methodology to our data. We realized that data distribution and the complexity of the dataset, with interacting factors, repeated measures within subjects, and across experimental protocols would not fulfill the assumptions required to perform such testing and would further not result in a comprehensive account of the results. Instead, we decided to fit GLMMs to the data that allow for an easier and more comprehensive description. References J. Driver, A selective review of selective attention research from the past century. British J of Psychology 92 , 53–78 (2001). M. Domjan, J. W. Grau, Principles of learning and behavior , 7th ed (Cengage Learning, 2015). J. H. Byrne, Ed., Learning and memory: a comprehensive reference , 1st ed (Elsevier, 2008). M.-A. Mackie, N. T. Van Dam, J. Fan, Cognitive control and attentional functions. Brain and Cognition 82 , 301–312 (2013). G. R. Esber, M. Haselgrove, Reconciling the influence of predictiveness and uncertainty on stimulus salience: a model of attention in associative learning. Proc. R. Soc. B. 278 , 2553–2561 (2011). P. Balsam, M. Drew, C. Gallistel, Time and Associative Learning. CCBR 5 , 1–22 (2010). H. Lachnit, “The Principle of Contiguity” in Principles of Learning and Memory , R. H. Kluwe, G. Lüer, F. Rösler, Eds. (Birkhäuser Basel, 2003), pp. 3–13. B. A. Williams, The critical dimensions of the response-reinforcer contingency. Behavioural Processes 54 , 111–126 (2001). A. E. Meuret, K. B. Wolitzky-Taylor, M. P. Twohig, M. G. Craske, Coping Skills and Exposure Therapy in Panic Disorder and Agoraphobia: Latest Advances and Future Directions. Behavior Therapy 43 , 271–284 (2012). M. G. Craske, M. Treanor, C. C. Conway, T. Zbozinek, B. Vervliet, Maximizing exposure therapy: An inhibitory learning approach. Behaviour Research and Therapy 58 , 10–23 (2014). F. Raeder, C. J. Merz, J. Margraf, A. Zlomuzica, The association between fear extinction, the ability to accomplish exposure and exposure therapy outcome in specific phobia. Sci Rep 10 , 4288 (2020). M. E. Bouton, N. E. Winterbauer, T. P. Todd, Relapse processes after the extinction of instrumental learning: Renewal, resurgence, and reacquisition. Behavioural Processes 90 , 130–141 (2012). T. P. Todd, D. Vurbic, M. E. Bouton, Mechanisms of renewal after the extinction of discriminated operant behavior. Journal of Experimental Psychology: Animal Learning and Cognition 40 , 355–368 (2014). M. E. Bouton, S. Maren, G. P. McNally, Behavioral and neurobiological mechanisms of pavlovian and instrumental extinction learning. Physiological Reviews 101 , 611–681 (2021). J. M. Rosas, T. P. Todd, M. E. Bouton, Context change and associative learning. WIRES Cognitive Science 4 , 237–244 (2013). J. W. Rudy, Context representations, context functions, and the parahippocampal-hippocampal system. Learn Mem 16 , 573–585 (2009). S. Maren, K. L. Phan, I. Liberzon, The contextual brain: implications for fear conditioning, extinction and psychopathology. Nat Rev Neurosci 14 , 417–428 (2013). M. S. Fanselow, E. Eich, S. M. Smith, M. E. Bouton, “Part 5. Context” in Science of Memory , 1st Ed., (Oxford University Press, 2007), pp. 99–121. T. Walther, et al. , Context-dependent extinction learning emerging from raw sensory inputs: a reinforcement learning approach. Sci Rep 11 , 2713 (2021). J. R. Donoso, et al. , Emergence of complex dynamics of choice due to repeated exposures to extinction learning. Anim Cogn 24 , 1279–1297 (2021). J. M. Peschken, L. Hahn, R. Pusch, J. Rose, Context is Learned, not Given. [Preprint] (2025). Available at: https://www.researchsquare.com/article/rs-5682968/v1 [Accessed 3 March 2025]. J. J. Medina Peschken, “Context as a learned property of stimuli in extinction learning” (Animal Study Registry, 2023). T. Otto, J. Rose, The open toolbox for behavioral research. Behav Res 56 , 4522–4529 (2023). D. H. Brainard, The Psychophysics Toolbox. Spatial Vis 10 , 433–436 (1997). J. R. Donoso, et al. , Emergence of complex dynamics of choice due to repeated exposures to extinction learning. Anim Cogn 24 , 1279–1297 (2021). J. Packheiser, O. Güntürkün, R. Pusch, Renewal of extinguished behavior in pigeons (Columba livia) does not require memory consolidation of acquisition or extinction in a free-operant appetitive conditioning paradigm. Behavioural Brain Research 370 , 111947 (2019). J. Packheiser, J. R. Donoso, S. Cheng, O. Güntürkün, R. Pusch, Trial-by-trial dynamics of reward prediction error-associated signals during extinction learning and renewal. Progress in Neurobiology 197 , 101901 (2021). M. E. Bouton, Context and behavioral processes in extinction. Learn Mem 11 , 485–494 (2004). B. L. Thomas, N. Larsen, J. J. B. Ayres, Role of context similarity in ABA, ABC, and AAB renewal paradigms: Implications for theories of renewal and for treating human phobias. Learning and Motivation 34 , 410–436 (2003). S. Y.-S. Khoo, J. M. Sciascia, A. Brown, N. Chaudhri, Comparing ABA, AAB, and ABC Renewal of Appetitive Pavlovian Conditioned Responding in Alcohol- and Sucrose-Trained Male Rats. Front. Behav. Neurosci. 14 , 5 (2020). W. Clark, M. Colombo, Seeing the Forest for the Trees, and the Ground Below My Beak: Global and Local Processing in the Pigeon’s Visual System. Front. Psychol. 13 , 888528 (2022). F. A. Soto, E. A. Wasserman, Mechanisms of object recognition: what we have learned from pigeons. Front. Neural Circuits 8 (2014). “The Avian Visual System: Overview” in How Animals See the WorldComparative Behavior, Biology, and Evolution of Vision , (Oxford University Press, 2012), pp. 473–482. M. A. McDevitt, E. Fantino, Visual stimulus compounding with pigeons. Behavioural Processes 38 , 265–275 (1996). K. Goto, S. Watanabe, The whole is equal to the sum of its parts: Pigeons (Columba livia) and crows (Corvus macrorhynchos) do not perceive emergent configurations. Learn Behav 48 , 53–65 (2020). D. I. Brooks, R. G. Cook, K. Goto, Perceptual grouping and detection of trial-unique emergent structures by pigeons. Anim Cogn 25 , 717–729 (2022). K. K. Cavoto, R. G. Cook, Cognitive precedence for local information in hierarchical stimulus processing by pigeons. Journal of Experimental Psychology: Animal Behavior Processes 27 , 3–16 (2001). J. Emmerton, J. C. Renner, Local rather than global processing of visual arrays in numerosity discrimination by pigeons (Columba livia). Anim Cogn 12 , 511–526 (2009). U. Aust, E. Braunöder, Transfer between local and global processing levels by pigeons (Columba livia) and humans (Homo sapiens) in exemplar- and rule-based categorization tasks. Journal of Comparative Psychology 129 , 1–16 (2015). K. Goto, A. J. Wills, S. E. G. Lea, Global-feature classification can be acquired more rapidly than local-feature classification in both humans and pigeons. Animal Cognition 7 , 109–113 (2004). Z. Rezvani, A. Katanforoush, H. Pouretemad, Global precedence changes by environment: A systematic review and meta-analysis on effect of perceptual field variables on global-local visual processing. Atten Percept Psychophys 82 , 2348–2359 (2020). S. P. León, A. M. Gámez, J. M. Rosas, Mechanisms of Contextual Control when Contexts are Informative to Solve the Task. Span. j. psychol. 15 , 10–19 (2012). S. Lucke, H. Lachnit, S. Koenig, M. Uengoer, The informational value of contexts affects context-dependent learning. Learn Behav 41 , 285–297 (2013). S. Lucke, H. Lachnit, M. C. Stüttgen, M. Uengoer, The impact of context relevance during extinction learning. Learn Behav 42 , 256–269 (2014). N. J. Mackintosh, An Analysis of Overshadowing and Blocking. Quarterly Journal of Experimental Psychology 23 , 118–125 (1971). N. J. Mackintosh, Overshadowing and stimulus intensity. Animal Learning & Behavior 4 , 186–192 (1976). C. Bonardi, E. Mondragón, B. Brilot, D. J. Jennings, Overshadowing by fixed- and variable-duration stimuli. Quarterly Journal of Experimental Psychology 68 , 523–542 (2015). D. J. Jennings, C. Bonardi, K. Kirkpatrick, Overshadowing and stimulus duration. Journal of Experimental Psychology: Animal Behavior Processes 33 , 464–475 (2007). K. J. Leising, D. Garlick, A. P. Blaisdell, Overshadowing between landmarks on the touchscreen and in arena with pigeons. Journal of Experimental Psychology: Animal Behavior Processes 37 , 488–494 (2011). M. R. Horne, J. M. Pearce, Potentiation and overshadowing between landmarks and environmental geometric cues. Learn Behav 39 , 371–382 (2011). M. L. Spetch, Overshadowing in landmark learning: Touch-screen studies with pigeons and humans. Journal of Experimental Psychology: Animal Behavior Processes 21 , 166–181 (1995). R. Deery, S. Commins, Landmark Distance Impacts the Overshadowing Effect in Spatial Learning Using a Virtual Water Maze Task with Healthy Adults. Brain Sciences 13 , 1287 (2023). J. M. Rosas, J. E. Callejas Aguilera, M. M. Ramos Álvarez, M. J. Fernández Abad, Revision of retrieval theory of forgetting: What does make information context-specific? International Journal of Psychology and Psychological Therapy 6 , 147–166 (2006). J. M. Rosas, J. E. Callejas-Aguilera, Context switch effects on acquisition and extinction in human predictive learning. Journal of Experimental Psychology: Learning, Memory, and Cognition 32 , 461–474 (2006). P. M. Ogallar, M. M. R. Álvarez, J. A. Alcalá, M. M. M. Fernández, J. M. Rosas, Attentional perspectives on context-dependence of information retrieval. International Journal of Psychology and Psychological Therapy 17 , 121–136 (2017). E. M. O’Donoghue, L. Castro, E. A. Wasserman, Hierarchical and configural control in conditional discrimination learning. Journal of Experimental Psychology: Animal Learning and Cognition 48 , 370–382 (2022). P. M. Blough, Attentional priming and visual search in pigeons. Journal of Experimental Psychology: Animal Behavior Processes 15 , 358–365 (1989). S. L. Gray, M. A. J. Qadri, D. I. Brooks, R. G. Cook, Use of different attentional strategies by pigeons and humans in multidimensional visual search. Journal of Experimental Psychology: Animal Learning and Cognition 48 , 46–59 (2022). Additional Declarations The authors declare no competing interests. Supplementary Files Supplementaryinformation.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6146552","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":423395490,"identity":"5a7b406e-65c1-43ce-b782-5573f857e90f","order_by":0,"name":"Juan Peschken","email":"data:image/png;base64,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","orcid":"https://orcid.org/0000-0001-5480-1653","institution":"Ruhr-University Bochum, Neural Basis of Learning","correspondingAuthor":true,"prefix":"","firstName":"Juan","middleName":"","lastName":"Peschken","suffix":""}],"badges":[],"createdAt":"2025-03-03 13:15:17","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":true,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":true},"doi":"10.21203/rs.3.rs-6146552/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6146552/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":78795475,"identity":"10e2ee39-cf69-4155-bdd9-629349ee4f2c","added_by":"auto","created_at":"2025-03-19 05:10:52","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":118399,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e(A) Experimental setup. The arena was illuminated with white LED lights, and large monitors displayed geometric shapes as landmarks at the ends of each arm. The animals responded and received rewards on small touchscreens close to the floor. (B) Trial structure. Following a successful initiation and a one-second delay, a forced choice between S+ and S- stimuli was presented. These stimuli were either familiar (consistent across all sessions) or novel (session-specific). The screen location and sequence of familiar and novel stimuli were pseudo-randomized. (C) Experimental context, ABA’ procedure. Acquisition (Phase A) took place in a randomly assigned location, with the touchscreen displaying a white background and the arm monitors showing the usual geometric shape on a black background. Extinction (Phase B) altered all three contextual stimuli: the animals moved to a different location (Spatial context), the touchscreen color changed to yellow (Local context), and the monitor color changed to red (Global context).\u003c/em\u003e \u003cem\u003eFinally, in the renewal test (Phase A'), examples of the condition for each context are shown. Only one of the contextual stimuli returns to the configuration of Phase A\u003c/em\u003e\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6146552/v1/b49d8e7ba61c3c276d992986.png"},{"id":78795119,"identity":"ee23c765-4be9-4889-85f7-03bbd7287be0","added_by":"auto","created_at":"2025-03-19 05:02:52","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":64948,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eModel 1 - Response patterns for familiar and novel stimuli across phases. Animals retained a high probability of CR for familiar stimuli across all three phases, while novel stimuli exhibited dynamic changes. During ACQ, discrimination for novel stimuli was successfully acquired within the first 20 trials. In EXT, CR responses began high but rapidly declined due to the absence reinforcement. At the onset of REN, an increased probability of CR for novel stimuli indicated the renewal effect. Dots represent individual data points, dotted gray line denotes criterion for novel in each phase.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-6146552/v1/7924ee62b15003c0a3f0205a.png"},{"id":78795499,"identity":"a1252c71-ba37-450b-ac8e-76c2b0edb589","added_by":"auto","created_at":"2025-03-19 05:10:52","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":64703,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eModel 2 - Renewal responses by context across sessions. The different contexts elicited renewal responses with varying strength. Although all contexts showed a decrease in responses over repeated protocol sessions (decay effect), the local context consistently led to the highest renewal, followed by the spatial context, while the global context produced comparatively minimal renewal. Dots represent individual data points.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-6146552/v1/6e8eba6f90a3849e23301af7.png"},{"id":78795117,"identity":"9b7f72a6-3e9c-4f3c-b843-363ce704d2eb","added_by":"auto","created_at":"2025-03-19 05:02:52","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":97326,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e(A) Model 3 - Latency across contexts. The local context exhibited the lowest latency (renewal onset), with the typical first renewal response occurring between trials 1-2 (M = 1.8). Responses in the spatial context started between trials 3-4 (M = 3.87), while renewal in the global context began much later, between trials 9-10 (M = 9.56). (B) Density distributions of latency across contexts. The different contexts showed distinct distribution shapes: the local context had a narrow range with a high peak around the second trial, the spatial context, while having a smaller latency peak, exhibited a much longer tail. In contrast, the global context displayed a peak at a later latency with a curved bell-shaped distribution.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-6146552/v1/bbbe5584bd802d17ef6d0436.png"},{"id":78795118,"identity":"4f8178d5-bc29-4bbb-b1ee-6d640ad1ef12","added_by":"auto","created_at":"2025-03-19 05:02:52","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":115842,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e(A) C1 – Local contexts across experiments. Increasing the continuity of the local context (local+) led to a higher number of renewal responses across all testing sessions. (B) C2 – Spatial contexts across experiments. Similarly, increasing the salience of the spatial context (spatial+) resulted in more renewal responses. Solid lines represent the main effect of each context across sessions, while dots indicate the actual number of renewal responses per session. Overall, the local context produced more renewal than the spatial context. Due to this difference, the axes do not share the same scale; relevant comparisons should be made within each panel separately.\u003c/em\u003e\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-6146552/v1/ab62a840dbe23faa12c2ef88.png"},{"id":78796970,"identity":"7fd8afd5-2cbd-4b0e-a28c-8294cad1065c","added_by":"auto","created_at":"2025-03-19 05:34:57","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":906251,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6146552/v1/63f60edb-3440-4d24-814a-78729d311226.pdf"},{"id":78795113,"identity":"e055151c-a82d-4520-bb21-f88d2efdc792","added_by":"auto","created_at":"2025-03-19 05:02:52","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":276742,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaryinformation.docx","url":"https://assets-eu.researchsquare.com/files/rs-6146552/v1/c803d1e185eb6675887b0750.docx"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003e\u003cstrong\u003eContext Emergence through Stimulus Competition in Extinction Learning\u003c/strong\u003e\u003c/p\u003e","fulltext":[{"header":"Introduction","content":"\u003cp\u003eIn the process of adapting to the environment, organisms are continuously exposed to a multitude of stimuli that compete for attention and relevance. Some of these stimuli are selected and prioritized, leading to learning processes in which certain stimuli are sought out for their beneficial outcomes, while others are avoided to prevent harmful consequences. But how does an organism determine which stimuli are useful and contain informative value?\u003c/p\u003e\n\u003cp\u003eThis selection mechanism has been studied extensively, leading to the development of numerous theories on learning and attention (1\u0026ndash;4). Today, we know that each organism may have a particular approach to determining this competition. However, in general, the selection of relevant stimuli depends on two main aspects: internal factors associated with the organism, such as sensory capabilities and the significance of each stimulus according to its biological needs, and external factors, related to the stimulus itself, including the relative salience compared to other stimuli (5), its duration and stability (continuity), (6) the temporal proximity to associated consequences (contiguity) (7), and the reliability of its occurrence with those consequences (contingency) (8).\u003c/p\u003e\n\u003cp\u003eIf these elements are relevant for learning the relationships between stimuli, behavior and outcomes, they should also play a crucial role when those relationships cease to function and an \u0026lsquo;update\u0026rsquo; is necessary. Central to this capacity is extinction learning, a process through which previously learned associations are gradually diminished in response to the omission of expected outcomes. However, this adjustment is not always straightforward. While extinction learning is crucial for the suppression of maladaptive behaviors, such as pathological fear responses or problematic conducts (9\u0026ndash;11), it does not necessarily result in a permanent loss of learned associations. Extinction learning is best understood as a new learning process that conditionally inhibits previous associations. Several of these conditions are evident in the phenomenon of renewal (12, 13), where extinguished behaviors reemerge when an individual changes the context that no longer supports the learned association.\u003c/p\u003e\n\u003cp\u003eThe renewal effect has been studied in various experimental paradigms, most commonly the ABA design (14), where animals first learn an association in context A, undergo extinction in context B, and are then tested in the original context A. Renewal is observed when the conditioned behavior reappears in response to the return to context A. While this classic framework has been widely used, the precise nature of context\u0026mdash;what stimuli are involved, how they interact, and how their influence on behavior is determined\u0026mdash;has traditionally been neglected. Context has often been defined as a passive background of environmental cues, such as spatial or sensory features that surround a learning experience (15\u0026ndash;18).\u003c/p\u003e\n\u003cp\u003eHowever, a growing body of research suggests that context is not merely a fixed environmental backdrop. Instead, context appears to be learned dynamically from stimuli present when the original association ceases to function (19\u0026ndash;21). Critically, when learning occurs among multiple individual stimuli, competition arises between them. This competitive view of context implies that no single stimulus inherently serves as the context, nor is context simply a compound of all available stimuli. Rather, it emerges from the relative influence of competing stimuli, each contending for control over the learned behavior.\u003c/p\u003e\n\u003cp\u003eTo further explore this framework, we designed a novel experimental approach in pigeons aimed at systematically assessing how different types of stimuli\u0026mdash;spatial, local, and global\u0026mdash;compete to serve as the context for extinction and renewal (22). During extinction in context B, all three types of stimuli were altered, but only one returned to its original form in the test context (A\u0026rsquo;) forming an ABA\u0026apos; paradigm. All stimuli were controlled to have the same level of contingency and contiguity, while specific parameters of continuity and saliency allowed for meaningful comparisons with previous reports. By manipulating these conditions, we aimed to determine which cues are more likely to be selected as the context and how their competitive dynamics influence the renewal of conditioned responses.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cp\u003e\u003cstrong\u003eSubjects.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe conducted the experiment on seven pigeons (\u003cem\u003eColumba livia\u003c/em\u003e), consisting of four females and three males of unknown age, sourced from local breeders and individually housed in a colony room. Throughout the experimental procedures, the birds had unrestricted access to water and were maintained on a controlled feeding protocol. They received food rewards during the experimental sessions. These pigeons are part of a group of animals that provided a complete dataset, as detailed in this study pre-registration (22). For this experiment, three animals were completely na\u0026iuml;ve, while four had previously participated in a similar protocol. There was at least a three-month gap between the two protocols. All experimental procedures complied with the German guidelines for the care and use of animals in science, as well as the European Communities Council Directive 86/609/EEC on the care and use of animals for experimental purposes. The study was approved by the ethics committee of the State of North Rhine Westphalia, Germany.\u003c/p\u003e\n\u003cp id=\"_Toc114688403\"\u003e\u003cstrong\u003eExperimental setup\u003c/strong\u003e\u003cstrong\u003e.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe experimental setup and initial training procedures were conducted as previously described in (21)\u003cstrong\u003e.\u0026nbsp;\u003c/strong\u003eBriefly, we used a plus-shaped arena (3.5m x 3.5m) composed of four identical arms, each illuminated by programmable LED lamps. At the end of each arm, a 43\u0026rdquo; monitor displayed geometrical shapes as landmarks. Overall, the arena contained eight distinctive locations where the birds could interact and perform the task. A camera was installed at the center of the arena\u0026rsquo;s ceiling to observe and record the birds\u0026rsquo; behavior. A water bowl was always present in the center of the arena at floor level. The layout provided eight distinct touchscreen locations for the birds to interact with and receive rewards, with one active location at a time. The open design allowed the pigeons to move freely, including flying in and out, mirroring their natural foraging behavior.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eThree contextual dimensions in the arena setup.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe configuration of the arena employed in this experiment allowed controlling context with three distinct stimuli (Fig. 1A).The \u003cem\u003espatial context\u003c/em\u003e, defined by the physical position of the animal in the arena, enriched by a large landmark displayed in the 43\u0026rdquo; monitor at the end of each arm, a geometrical figure (triangle, hexagon, horizontal bar, quarter circle), the landmarks were always displayed in the same monitors, remaining stable throughout the experiment. The \u003cem\u003elocal context,\u0026nbsp;\u003c/em\u003ewhich was the background color of the 7\u0026rdquo; active touchscreen on which cue-stimuli were displayed and selected. The colors followed the RGB color model, overall, six different colors were used, the main three colors and their most extreme combinations: Red (1, 0, 0), Green (0, 1, 0), Blue (0, 0, 1), Yellow (1, 1, 0), Cyan (0, 1, 1) and Magenta (1, 0, 1). The \u003cem\u003eglobal context\u003c/em\u003e, designated by the background color of the 43\u0026rdquo; monitor at the end of the arm where the task was being performed, four colors were used, Red (1, 0, 0), Green (0, 1, 0), Blue (0, 0, 1) and Yellow (1, 1, 0).\u003c/p\u003e\n\u003cp\u003eThe birds were randomly assigned to face a specific context in their first session and then continued alternating contexts in each subsequent session, each context was tested four times. Overall, the seven pigeons underwent a total of 84 sessions (12 each). The capacity of each stimulus to function as a context was evaluated by counting the number of renewal responses when the respective context returned to the acquisition array after extinction. Custom MATLAB code (Mathworks, Natick, MA, USA) was used to control stimulus presentation and contextual features, utilizing the open toolbox for behavioral research (23) and the Psychophysics toolbox (24).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBehavioral protocol.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe pigeons performed a forced choice discrimination task, originally adapted from (25\u0026ndash;27) and similar to the employed in (21) (Fig. 1B). In each session, we used two pairs of choice stimuli (S+ and S-), one familiar and one novel. The familiar pair was pre-trained and remained constant across all sessions, serving as a control and maintaining task engagement throughout the different phases of a session. Conditioned responses to the familiar stimuli (S+) were always rewarded, enabling us to control satiety levels and potential fatigue during the session. The novel pair, unique to each session and never presented before, were the critical stimuli for testing acquisition, extinction, and renewal. Both familiar and novel stimuli, as well as their screen locations, were presented in a pseudo-randomized and balanced order.\u003c/p\u003e\n\u003cp\u003eThe session began with the pigeon flying freely into the arena, searching for the active touchscreen to initiate the task. Each trial began with the presentation of an initiation stimulus, a black dot centered on a touchscreen, displayed for a maximum of ten seconds. The trial was initiated by a single peck to this stimulus, followed by a one-second delay before the choice stimuli appeared on the screen for up to six seconds. Three possible interactions could occur: a peck on the S+ (\u0026lsquo;correct response\u0026rsquo;) resulted in a reward of one food pellet and the illumination of the LED below the feeder for 0.5 seconds; a peck on the S- (\u0026lsquo;alternative response\u0026rsquo;) resulted in a black screen for one second, no food reward, and an additional one-second delay before the next trial began; no interaction with either stimulus during the presentation period (\u0026lsquo;no response\u0026rsquo;) resulted in no food reward or additional feedback. Consecutive trials were separated by an inter-trial interval (ITI) of four seconds. Once all trials at a location were completed, a black-and-white checkerboard appeared, signaling the animal that the current touchscreen was inactive, encouraging it to search for another site\u003c/p\u003e\n\u003cp\u003eThree distinct phases, acquisition (A), extinction (B), and renewal (A) took place within each session, all with a specific contextual array including all three contextual stimuli (Fig. 1C), effectively establishing an ABA protocol (28, 29).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eDuring the acquisition phase, subjects learned the S+/S\u0026minus; association for the novel stimuli through trial and error. The physical array of the arena was constant during each acquisition session (A). \u0026nbsp;For the local context, all touchscreen backgrounds were white, while for the global context, all monitor backgrounds were black. For the spatial context, to ensure balanced exposure across sessions, we predetermined a random order of the eight possible locations. This order was followed for all subsequent sessions, ensuring that after completing the full cycle of eight locations, the repeated spatial context was temporally distant from the previous first exposure. Acquisition consisted of a minimum of 80 trials (40 for each pair of stimuli) and continued until the subject met two criteria: completing the minimum trial count and reaching 85% correct responses in the last 20 trials for both familiar and novel stimuli. After the acquisition phase was completed, a one-minute ITI was implemented. The bird then moved freely, looking for the next active touchscreen to continue the task in the extinction phase.\u003c/p\u003e\n\u003cp\u003eThe extinction phase followed in a new contextual configuration, altering all three context stimuli. Animals performed the task on a different touchscreen located in a separate arm of the arena (spatial context), with the touchscreen background changing to a new color (local context). This change was triggered only after the trial was successfully initiated, the birds began the trial with the known white background, and the color change occurred only after initiation, in conjunction with the appearance of the choice stimuli (S+ and S-).The monitor background in the new arm also changed to a different color (global context), this change was stable thought the extinction phase. During this phase, responses to the novel stimuli no longer elicited feedback (neither food nor timeout); instead, interactions with S+ and S\u0026minus; simply cleared the screen, returning it to the previous white background and initiated the standard ITI. The extinction phase consisted of a minimum of 80 trials (40 familiar, 40 novel) and concluded only when subjects met two criteria: completing the minimum trial count and achieving 85% correct responses on familiar stimuli and 85% extinction on novel stimuli, both calculated over the last 20 trials. Extinction was defined as trials where the bird either responded to the S\u0026minus; or omitted a response entirely, while responses to S+ were classified as a failure to extinguish. Once the extinction phase was complete, a second one-minute ITI followed. Here, the animal roamed through the arena in search of the final active screen.\u003c/p\u003e\n\u003cp\u003eIn the renewal phase, only one of the three contextual stimuli reverted to its original acquisition configuration, while the remaining stimuli stayed as they were during extinction (A\u0026rsquo;). This setup created a \u0026lsquo;competition\u0026rsquo; among the contextual stimuli, enabling a direct comparison of renewal effects across different context types. During renewal, responses to the novel stimuli continued without feedback, mirroring extinction, to observe the renewal effect. Renewal was induced one minute later, controlling for potential recovery effects like reinstatement and spontaneous recovery, ensuring the observed responses were due to the renewal effect. The renewal phase comprised a minimum of 40 trials (20 familiar, 20 novel) and concluded only when the subject met two criteria: completing the minimum trial count and achieving 85% extinction responses in the last 20 trials, defined as the absence of responses to the unrewarded S+.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistical analysis. \u0026nbsp;\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe sample size for this study was determined through a priori power analysis using G*Power 3.1 software (24). Based on a preliminary pilot study, we assumed an effect size of dz = 1.09, a significance level of \u0026alpha; = 0.05, and a power of 0.8. Data analysis was performed using custom MATLAB code (25), while GLMM-specific analyses were conducted in R Statistical Software (26) with the packages lme4 (27), flexplot (28), and DHARMa (29).\u003c/p\u003e\n\u003cp\u003eTo assess successful acquisition, extinction, and renewal, we recorded the number of responses and compared them to pre-defined criteria: 17 out of 20 correct responses (i.e., 85%) in the acquisition phase, and 17 out of 20 extinction responses (either no response or alternative responses) in the extinction and renewal phases. This criterion was considered statistically significant (p \u0026lt; 0.05) under a cumulative probability within the binomial distribution. The probability of obtaining at least 17 correct responses out of 20 trials by chance, assuming a 50% random choice, was calculated to be 0.00020123.\u003c/p\u003e\n\u003cp\u003eWe used multiple generalized linear mixed models (GLMMs) to analyze the response patterns to different types of stimuli across experimental phases and to evaluate the strength of the renewal effect for each contextual stimulus. In total, three distinct models were fitted. The first model examined the change of familiar and novel stimuli across the different phases. For this purpose, a binary outcome variable (Response: correct = 1, incorrect = 0) was modeled based on three main fixed effects and their three-way interaction. These fixed factors included experimental phase (Phase: acquisition, extinction, and renewal), target stimulus (Stimulus: familiar and novel), and trial within phase (Trial: trial count for each Stimulus within each Phase). The random-effects structure of the model included Sessions nested within Subjects. To account for the variable session lengths in our design, the specific interaction between Stimulus and Trial was incorporated as a random term.\u0026nbsp;The model was fitted using a Binomial distribution and the logit link-function.\u003c/p\u003e\n\u003cp\u003eM1 = Response ~ Stimuli*Phase*Trial + (Stimuli:Trial | Subject:Session)\u003c/p\u003e\n\u003cp\u003eIn the second model, the number of renewal responses to the novel stimuli was analyzed based on two main fixed factors. These factors were the contextual stimulus (Context: Global, Spatial, and Local) and the contextual session number, representing each test within a specific context (CtxtSess: 1, 2, 3, and 4). The random-effects structure included Sessions nested within Subjects. The model was fitted using a Poisson distribution and the log link-function.\u003c/p\u003e\n\u003cp\u003eM2 = Renewal ~ Context + CtxtSession + (1 | Subject:Session)\u003c/p\u003e\n\u003cp\u003eThe third and final model aimed to analyze the temporal dynamics of renewal onset, considering only sessions where renewal responses were observed. Sessions that did not produce renewal were excluded from this analysis. For this purpose, we established a latency metric, defined as the number of trials required to initiate renewal responses. This was then analyzed as a function of two main factors: Context (Global, Spatial, and Local) and the amount of renewal in each session. To better capture the influence of each context on latency, Context was included as a random factor. The random-effects structure likewise accounted for sessions nested within subjects. Since the model dealt with count data, it was fitted using a Poisson distribution with a log link function.\u003c/p\u003e\n\u003cp\u003eM3 = Latency ~ Context + Renewal + (Context | Subject:Session)\u003c/p\u003e\n\u003cp\u003eIn addition to the analysis performed on the data acquired in this experiment, two additional contrasts were made with previously published data (21). These comparisons, (refer to supplementary material SI1 \u0026amp; SI2) examined the influence of an increase in continuity for the local context (C1) and an increase in saliency for the spatial context (C2) in the amount of renewal responses. Each context was tested four times per animal, with the same seven animals participating across studies, resulting in a total of 56 sessions in each comparison\u0026mdash;28 for each context. The analysis described here is novel and represents the first time these specific comparisons have been conducted.\u003c/p\u003e\n\u003cp\u003eIn C1, the effect of increased continuity in the local context (local+) was tested by extending the background color change on the touchscreen, which was not only present during the choice stimulus (as implemented in this study) but also during the presentation of the initiation stimulus. On the other hand, C2 assessed the effect of enhanced saliency for the spatial context (spatial+). In this comparison, in addition to the different locations enabled by the arena, spatial information was enhanced by integrating the global context, using monitors with a constant color that corresponded to the illumination in each arm of the arena, making the different arms more recognizable. Both comparisons employed the same model structure where renewal responses were fitted as a function of two main factors, the contextual stimulus (Context: Local and Local+ for C1, or Spatial and Spatial+ for C2) and\u0026nbsp;contextual session number, counting each test within a specific context (CtxtSess: 1, 2, 3, and 4). The random-effects structure included Subjects to accommodate for the repeated measurements design. Both models were fitted using a Poisson distribution and the log link-function.\u003c/p\u003e\n\u003cp\u003eC = Renewal ~ Context + CtxtSession + (1 | Subject)\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eWe observed the behavioral responses of seven pigeons in a two-stimulus discrimination task using a within-session acquisition-extinction-renewal paradigm (ABA) in an open arena. Three distinct types of contextual stimuli were implemented: spatial context, local context (defined by the background color of the active touchscreen), and global context (designated by the background color of the large monitor at the end of the active arm in the arena). To investigate the contextual control of individual stimuli, only one of the three contextual stimuli reverted to its acquisition state during the renewal test (ABA\u0026rsquo;).\u003c/p\u003e \u003cp\u003eModel 1 explored the dynamic of conditioned response of familiar and novel stimuli across the three distinctive phases (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Familiar stimuli had a high initial probability of CR in ACQ, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\widehat{P}\\)\u003c/span\u003e\u003c/span\u003e = 0.96 (β\u0026thinsp;=\u0026thinsp;3.415, t= 15.401, p \u0026lt; 0.001) and remained unaffected across the change of phases, for EXT (β\u0026thinsp;=\u0026thinsp;0.482, t= 1.618, p = 0.105) and for REN (β = -0.007, t= -0.026, p\u0026thinsp;=\u0026thinsp;0.979). Trial had a marginal positive effect for familiar only in ACQ, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\Delta\\:}\\widehat{P}\\)\u003c/span\u003e\u003c/span\u003e = + 0.0008, (β\u0026thinsp;=\u0026thinsp;0.026, t= 2.719, p\u0026thinsp;=\u0026thinsp;0.006), and no significant deviations in EXT and REN. In contrast, the novel stimuli had a lower initial probability of CR in ACQ, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\widehat{P}\\)\u003c/span\u003e\u003c/span\u003e = 0.63, (β = -2.863, t= -12.503, p \u0026lt; 0.001) but a stronger positive effect of trial representing the expected learning curve, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\Delta\\:}\\widehat{P}\\)\u003c/span\u003e\u003c/span\u003e = + 0.034, (β\u0026thinsp;=\u0026thinsp;0.126, t= 8.389, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). As anticipated, trial had a negative effect for novel in EXT, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\Delta\\:}\\widehat{P}\\)\u003c/span\u003e\u003c/span\u003e\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;0.038, (β = -0.319, t= -17.814, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and REN, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\Delta\\:}\\widehat{P}\\)\u003c/span\u003e\u003c/span\u003e\u0026thinsp;=\u0026thinsp;\u0026minus;\u0026thinsp;0.030, (β = -0.277, t= -12.06, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), demonstrating the decay of performance due to violation of reward expectancy both in extinction and renewal.\u003c/p\u003e \u003cp\u003eModel 2 examined the influence of individual contextual stimuli on the return of conditioned responses, reflecting the strength of the renewal effect (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). All three contextual stimuli elicited renewal, albeit to different degrees. Without the effect of session, the global context yielded 2.7 responses as a pure effect (β\u0026thinsp;=\u0026thinsp;1.021, t\u0026thinsp;=\u0026thinsp;3.427, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), the spatial context generated 13.9 responses (β\u0026thinsp;=\u0026thinsp;1.614, t\u0026thinsp;=\u0026thinsp;5.447, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and, unexpectedly, the local context produced the highest response count of 29.2 (β\u0026thinsp;=\u0026thinsp;2.354, t\u0026thinsp;=\u0026thinsp;8.19, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). There was a significant decay in renewal for all contexts across repeated tests, with an average reduction of 48.5% of responses per session (β = -0.663, t = -7.491, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e \u003cp\u003eWhen analyzing the temporal dynamics of renewal, Model 3 identified differences in the typical latency across contexts (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA). Animals exhibited the shortest latency in the local condition, specifically when the local context matched the one from the acquisition phase, with an average renewal onset at trial 1.89 (β\u0026thinsp;=\u0026thinsp;0.638, t\u0026thinsp;=\u0026thinsp;2.590, p\u0026thinsp;=\u0026thinsp;0.012). In the spatial condition, renewal onset was consistently later, with an average latency of 3.87 trials (β\u0026thinsp;=\u0026thinsp;0.717, t\u0026thinsp;=\u0026thinsp;3.104, p\u0026thinsp;=\u0026thinsp;0.003). The global condition showed the longest latency, with renewal beginning on average at trial 9.57 (β\u0026thinsp;=\u0026thinsp;1.620, t\u0026thinsp;=\u0026thinsp;5.613, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Interestingly, the model revealed that the renewal predictor showed no relationship between the number of renewal responses in a session and latency, indicating that the \"intensity\" of renewal for each context was independent of when responses began (β\u0026thinsp;\u0026lt;\u0026thinsp;0.001, t\u0026thinsp;=\u0026thinsp;0.041, p\u0026thinsp;=\u0026thinsp;0.967).\u003c/p\u003e \u003cp\u003eThis finding sparked a new analysis. Intuitively, one might expect that if an animal is \u0026lsquo;highly certain\u0026rsquo; that the context has changed and that the reinforcer will become available again, renewal responses should emerge quickly and be more robust. Conversely, \u0026lsquo;lower certainty\u0026rsquo; about the context change should result in slower and weaker renewal responses. However, this was not the case in our data. While context modulated both the number of renewal responses and their typical onset, there was no relationship between response intensity (i.e., the duration or total number of responses) and latency.\u003c/p\u003e \u003cp\u003eRemaining agnostic about the underlying mechanisms driving this behavior, but with the intention of gaining more insight into this effect, we applied a Kernel smoothing function to the latency data to estimate the potential distribution of latencies across contexts (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB). To assess whether the resulting distributions differed significantly, we first simulated 1,000 data points from a linearly spaced vector spanning the minimum and maximum latency values and interpolated them to ensure that the resulting Kernel distributions were assessed on the same data pool. We then performed pairwise comparisons using a Bonferroni-corrected two-sample Kolmogorov-Smirnov test. The corrected p-value threshold for significance was set at 0.016. All three distributions were significantly different from each other: Local to Spatial (D\u0026thinsp;=\u0026thinsp;0.721, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), Local to Global (D\u0026thinsp;=\u0026thinsp;0.712, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and Spatial to Global (D\u0026thinsp;=\u0026thinsp;0.309, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). This indicates that not only do the contexts differ in the typical renewal onset latency, but also in the range of trials in which renewal responses are expected to occur.\u003c/p\u003e \n\u003cp\u003eWhen comparing results across studies, we found that increasing continuity (i.e., the duration of stimulus presentation) for the local context led to a higher number of renewal responses (Fig. 5A). In the first session the local+ produced 25.2 renewal responses (\u0026Delta; + 12.1 to local) (\u0026beta; = 0.660, t = 7.684, p \u0026lt; 0.001). A similar effect was observed for the increased salience in the spatial context, where spatial+ produced 13.2 renewal responses (\u0026Delta; + 6.1 to spatial) (\u0026beta; = 0. 619, t = 5.145, p \u0026lt; 0.001). These results reveal a clear trend: increasing the parameters that enhance stimulus learning leads to a higher number of renewal responses. Additionally, both comparisons also confirmed the expected decay effect with repeated testing: the number of renewal responses decreased as the number of test sessions increased. Specifically, the decay rate per session was 39.6% in each session for the local context (\u0026beta; = -0.505, t = -12.499, p \u0026lt; 0.001) and 46% for the spatial (\u0026beta; = -0.616, t = -10.329, p \u0026lt; 0.001).\u0026nbsp;\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, we expand on our new experimental approach, ABA\u0026apos;, focusing on the competition between potential contextual stimuli during extinction learning. This method enables direct testing of each context in competition with others, using the frequency of renewal responses as a proxy to assess how effectively a specific stimulus functions as a context. Three types of stimuli were employed in this experiment: global (large distal visual cues), spatial (continuous visuospatial surroundings), and local (small proximal visual cues). All three types produced renewal, though in a differential manner, both in terms of the intensity of renewal responses and the temporal dynamics of when renewal occurred. Comparisons between the current data and previous results confirmed that manipulating learning parameters, such as salience and continuity, modulates the strength of renewal responses.\u003c/p\u003e\n\u003cp\u003eContextual variables are often studied in isolation, with a single manipulated stimulus serving as the context, or all contextual stimuli returning to the acquisition state simultaneously. This simplifies the task for subjects, as they can easily detect a clear difference between the acquisition, extinction, and renewal phases. However, this can obscure the relative effectiveness of individual stimuli as context and their influence on behavior. A small, but noticeable, change in internal states or the environment can lead to renewal effects (29, 30). In contrast, our approach directly tests how a stimulus serves as the context by creating competition among the available stimuli. But how can we then explain the results we obtained?\u003c/p\u003e\n\u003cp\u003eFirst, it is important to emphasize that our design and the open nature of the arena ensure that the pigeon samples all available information and cannot avoid or \u0026quot;miss\u0026quot; any source. The pigeon moves freely within the arena, gaining access to both spatial and global information as it transitions between phases (acquisition, extinction, and renewal) and forages for the new location where the task takes place. Pigeons\u0026apos; visual and attentional capacities have been extensively studied, showing that they can simultaneously process multiple points within their visual field, even employing differential neural pathways. They exhibit simultaneous discrimination of frontal objects (tectofugal pathway) and lateral, more distant information (thalamofugal pathway) (31\u0026ndash;33), which aligns precisely with the placement of our local and global contextual cues. As a result, when providing a response, the pigeon is simultaneously exposed to all three sources of contextual information, which would suggest that the observed differences in the contextual renewal are due to either, the \u0026lsquo;properties\u0026rsquo; of the pigeons or the \u0026lsquo;properties\u0026rsquo; of the stimuli.\u003c/p\u003e\n\u003cp\u003eSecond, our results further demonstrate how the different information presented in the context was not sampled as a passive backdrop where extinction learning occurred, as would align with the classical definition of context. Instead, our pigeons showed differential responses to each stimulus, demonstrating that each one was processed and attended to in a distinct manner. This reinforces our proposal that the \u0026lsquo;formation\u0026rsquo; of context in extinction learning is an active effort by the subject to disambiguate the conflict of performing the same behavior while receiving different outcomes, context is the result of an active learning process, rather than just a passive backdrop of the environment. This is consistent with previous reports highlighting pigeons\u0026apos; tendency to treat different elements not as a compound, but as separate parts, processing each source of information individually (34\u0026ndash;36).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIt is conceivable that, given the evolutionary history of the pigeon and its perceptual apparatus, there may be an intrinsic hierarchy that explains why local, spatial, and global stimuli generate differential renewal, a so-called \u0026lsquo;pigeon property\u0026rsquo;. For example, in visual discrimination and search tasks, it has been shown that pigeons process information differently from primates, being more sensitive to local rather than global stimulus features (37\u0026ndash;39). However, studies have shown that this tendency is not static and can reverse, resulting in a preference for global elements, particularly when the exposure time to the relevant stimuli is extended (40, 41). In our paradigm, spatial and global information is presented continuously throughout the entire experimental session, unlike the local information, which is presented briefly and intermittently only during specific moments in the task. Considering the differential presentation of stimuli, the local context should be the most affected by the brief presentation times. This is further supported by the comparison with greater continuity: longer presentation times lead to a stronger response. These challenges the idea that the local context is stronger simply because it is presented for a shorter duration, as, in fact, the increased exposure appears to generate a greater response.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAlthough we cannot rule out the possibility that some intrinsic \u0026lsquo;pigeon property\u0026rsquo; contributes to the pattern of results we observed, we believe there are more compelling arguments supporting the role of \u0026lsquo;stimulus properties\u0026rsquo;. For instance, we have shown that the effectiveness of contexts can be modulated by the properties of the stimuli. Both local and spatial stimuli performed better when we increased their continuity and salience, respectively. Likewise, this modulation does not only go in one direction, when we reduced the contiguity of the local context, we disrupted its ability to function as a context, causing it to lose the contextual competition and fail to produce renewal (21). Additionally, other accounts have highlighted how task-related factors can influence contextual processing, especially if the context proves to be useful for task solving (42\u0026ndash;44). Although, by definition, none of our contexts directly assist in solving the discrimination between the stimuli providing food to the pigeon, we cannot deny that the proximity of the local context to the task stimuli might increase its perceived \u0026apos;relevance\u0026apos;, as it is the only context that occurs contingently with the task.\u003c/p\u003e\n\u003cp\u003eThis aligns with similar findings in the study of the overshadowing phenomenon. Overshadowing is a classical phenomenon in conditioning and learning, where one stimulus becomes more influential in the learning process, overshadowing the effect of another stimulus (45, 46). This occurs when two stimuli are presented together during conditioning, but one stimulus is more salient or noticeable, causing it to dominate the learning process. As a result, the less salient stimulus fails to acquire as much associative strength or influence. There have been attempts to modulate overshadowing through stimulus duration, but these efforts have yielded conflicting results (47) or have not been successful altogether (48). In contrast, other studies examining distance and proximity variables have shown that the spatial distance between stimuli can result in one stimulus becoming more salient than the other (49, 50), This has been extended to human participants as well as pigeons (51, 52) . This aligns with our findings, where it appears more likely that the observed hierarchy is not caused by temporal dynamics related to the presentation differences of the stimuli, but rather by the proximity of the different sources of information to the pigeon.\u003c/p\u003e\n\u003cp\u003eIf we have a hierarchy based on the \u0026lsquo;stimulus properties\u0026rsquo;, it makes sense to expect that the pigeons would follow this hierarchy when faced with a context shift and the potential return of the reward. This is precisely what we observe: the three distinct contexts show differentiated latency and exhibit a staircase-like behavior. They seem to first sample the local context, followed by the spatial context, and finally the global context. Only once they reach the context that has changed do they begin their renewal responses. Pigeons seem to process one stimulus before moving on to the next, focusing on one context at a time in a predetermined manner.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIf this were the case, the hierarchy would also explain the differential intensity of renewal. Upon returning to the renewal phase for the local context, the pigeon experiences the local change immediately; both novel and familiar trials indicate that the context has changed and that the reward should now be back.\u0026nbsp;However, if it is the spatial context that changes, meaning the pigeon actively moves to the original location, the local information is again the first to be sampled before the spatial one. The pigeon would then experience trials, both novel and familiar, indicating that it remains in the same local context of extinction, that nothing has changed, accumulating evidence that the reward has not returned and permeating the spatial context with this update.\u0026nbsp;Consequently, by the time the pigeon samples the global context information, it has already gone through all previous trials where both local and spatial contexts, indicate that nothing has changed and that it continues in extinction. If the global context does exhibit any renewal response, it arrives late and lasts for a short duration.\u003c/p\u003e\n\u003cp\u003eThe key to understanding this staircase-like dynamic may lie in how attention is allocated across these different contexts. Organisms generally disregard contextual information when it is not relevant\u0026mdash;such as when the task is clear and provides all the required details to obtain food. However, during extinction, when the outcome is omitted, organisms redirect their focus to the context, resulting in context-dependent processing (15, 53). It has been suggested that attention could guide the process of information search with the aim of disambiguating the absence of the expected outcome in extinction learning (54, 55). Additionally, there are references showing how contextual information can hierarchically organize pigeons\u0026apos; behavior (56). While other studies suggest that pigeons can learn from past experiences and become primed to allocate attention in a specific way (57). Furthermore, pigeons tend to be much more exploratory and prone to attentional switching in complex choice situations than humans (58). All of this leads us to consider that differences in stimulus properties and patterns of attention allocation and processing during context formation, could generate reasonably repeatable and predictable structures when selecting which stimuli can become the context in extinction learning.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThis has relevant implications not only for experimental work and basic science but also for translational efforts and clinical applications. If we could take these dynamics into account when designing experiments, we could have more control over the variables we want to be attended to and identify potential confounds in our experiments. Extending this to therapeutic applications, we could design environments where therapy is conducted to promote certain stimuli, and not others, to be selected as the context for extinction. This would be immensely useful in ensuring that after extinction, there is no relapse and reappearance of undesired effects. We could select a context that is easily generalizable to other situations, facilitating the retention of learned inhibition in various contexts. Alternatively, we could consider a \u0026lsquo;portable\u0026rsquo; stimulus or context that can be carried by the patient and used during more challenging stages of exposure, thereby preventing relapse during the process. Achieving this requires meticulous control of the variables described in this study and others of great importance, such as salience. Additionally, validating attentional processes with gaze tracking or neural correlates would be imperative to truly understand how the brain resolves the competition between different contexts.\u003c/p\u003e\n\u003cp\u003eIn conclusion, our study provides compelling evidence that contextual stimuli compete for attention and influence renewal responses in pigeons. The differential renewal observed across local, spatial, and global contexts suggests that pigeons actively process and prioritize contextual information based on stimulus properties, providing more evidence that context is not merely a backdrop for behavior and a passive element in extinction learning, but rather an active process carried out by the organism. This process can be modulated and should be understood to improve our understanding of extinction and limit the occurrence of relapse.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePlanned study protocol.\u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was part of a planned experimental protocol that was registered with animalstudyregistry.org, on 25th January 2023. Details of the planned protocol are available on the Animal study registry platform\u0026rsquo;s website at (22) https://www.animalstudyregistry.org/10.17590/asr.0000305.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe original protocol consisted of three experimental manipulations (E1 \u0026ndash; E3). This report covers the results of experimental manipulation E1. Results of experimental manipulation E2 and E3 can be found on Peschken et al., 2025 (4).\u003c/p\u003e\n\u003cp\u003eWe deviated from the original protocol with regard to the statistical analysis. We did not apply t-test and ANOVA methodology to our data. We realized that data distribution and the complexity of the dataset, with interacting factors, repeated measures within subjects, and across experimental protocols would not fulfill the assumptions required to perform such testing and would further not result in a comprehensive account of the results. Instead, we decided to fit GLMMs to the data that allow for an easier and more comprehensive description.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eJ. Driver, A selective review of selective attention research from the past century. \u003cem\u003eBritish J of Psychology\u003c/em\u003e \u003cstrong\u003e92\u003c/strong\u003e, 53\u0026ndash;78 (2001).\u003c/li\u003e\n\u003cli\u003eM. Domjan, J. W. Grau, \u003cem\u003ePrinciples of learning and behavior\u003c/em\u003e, 7th ed (Cengage Learning, 2015).\u003c/li\u003e\n\u003cli\u003eJ. H. Byrne, Ed., \u003cem\u003eLearning and memory: a comprehensive reference\u003c/em\u003e, 1st ed (Elsevier, 2008).\u003c/li\u003e\n\u003cli\u003eM.-A. Mackie, N. T. Van Dam, J. Fan, Cognitive control and attentional functions. \u003cem\u003eBrain and Cognition\u003c/em\u003e \u003cstrong\u003e82\u003c/strong\u003e, 301\u0026ndash;312 (2013).\u003c/li\u003e\n\u003cli\u003eG. R. Esber, M. Haselgrove, Reconciling the influence of predictiveness and uncertainty on stimulus salience: a model of attention in associative learning. \u003cem\u003eProc. R. Soc. B.\u003c/em\u003e \u003cstrong\u003e278\u003c/strong\u003e, 2553\u0026ndash;2561 (2011).\u003c/li\u003e\n\u003cli\u003eP. Balsam, M. Drew, C. Gallistel, Time and Associative Learning. \u003cem\u003eCCBR\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 1\u0026ndash;22 (2010).\u003c/li\u003e\n\u003cli\u003eH. Lachnit, \u0026ldquo;The Principle of Contiguity\u0026rdquo; in \u003cem\u003ePrinciples of Learning and Memory\u003c/em\u003e, R. H. Kluwe, G. L\u0026uuml;er, F. R\u0026ouml;sler, Eds. (Birkh\u0026auml;user Basel, 2003), pp. 3\u0026ndash;13.\u003c/li\u003e\n\u003cli\u003eB. A. Williams, The critical dimensions of the response-reinforcer contingency. \u003cem\u003eBehavioural Processes\u003c/em\u003e \u003cstrong\u003e54\u003c/strong\u003e, 111\u0026ndash;126 (2001).\u003c/li\u003e\n\u003cli\u003eA. E. Meuret, K. B. Wolitzky-Taylor, M. P. Twohig, M. G. Craske, Coping Skills and Exposure Therapy in Panic Disorder and Agoraphobia: Latest Advances and Future Directions. \u003cem\u003eBehavior Therapy\u003c/em\u003e \u003cstrong\u003e43\u003c/strong\u003e, 271\u0026ndash;284 (2012).\u003c/li\u003e\n\u003cli\u003eM. G. Craske, M. Treanor, C. C. Conway, T. Zbozinek, B. Vervliet, Maximizing exposure therapy: An inhibitory learning approach. \u003cem\u003eBehaviour Research and Therapy\u003c/em\u003e \u003cstrong\u003e58\u003c/strong\u003e, 10\u0026ndash;23 (2014).\u003c/li\u003e\n\u003cli\u003eF. Raeder, C. J. Merz, J. Margraf, A. Zlomuzica, The association between fear extinction, the ability to accomplish exposure and exposure therapy outcome in specific phobia. \u003cem\u003eSci Rep\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 4288 (2020).\u003c/li\u003e\n\u003cli\u003eM. E. Bouton, N. E. Winterbauer, T. P. Todd, Relapse processes after the extinction of instrumental learning: Renewal, resurgence, and reacquisition. \u003cem\u003eBehavioural Processes\u003c/em\u003e \u003cstrong\u003e90\u003c/strong\u003e, 130\u0026ndash;141 (2012).\u003c/li\u003e\n\u003cli\u003eT. P. Todd, D. Vurbic, M. E. Bouton, Mechanisms of renewal after the extinction of discriminated operant behavior. \u003cem\u003eJournal of Experimental Psychology: Animal Learning and Cognition\u003c/em\u003e \u003cstrong\u003e40\u003c/strong\u003e, 355\u0026ndash;368 (2014).\u003c/li\u003e\n\u003cli\u003eM. E. Bouton, S. Maren, G. P. McNally, Behavioral and neurobiological mechanisms of pavlovian and instrumental extinction learning. \u003cem\u003ePhysiological Reviews\u003c/em\u003e \u003cstrong\u003e101\u003c/strong\u003e, 611\u0026ndash;681 (2021).\u003c/li\u003e\n\u003cli\u003eJ. M. Rosas, T. P. Todd, M. E. Bouton, Context change and associative learning. \u003cem\u003eWIRES Cognitive Science\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, 237\u0026ndash;244 (2013).\u003c/li\u003e\n\u003cli\u003eJ. W. Rudy, Context representations, context functions, and the parahippocampal-hippocampal system. \u003cem\u003eLearn Mem\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 573\u0026ndash;585 (2009).\u003c/li\u003e\n\u003cli\u003eS. Maren, K. L. Phan, I. Liberzon, The contextual brain: implications for fear conditioning, extinction and psychopathology. \u003cem\u003eNat Rev Neurosci\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 417\u0026ndash;428 (2013).\u003c/li\u003e\n\u003cli\u003eM. S. Fanselow, E. Eich, S. M. Smith, M. E. Bouton, \u0026ldquo;Part 5. Context\u0026rdquo; in \u003cem\u003eScience of Memory\u003c/em\u003e, 1st Ed., (Oxford University Press, 2007), pp. 99\u0026ndash;121.\u003c/li\u003e\n\u003cli\u003eT. Walther, \u003cem\u003eet al.\u003c/em\u003e, Context-dependent extinction learning emerging from raw sensory inputs: a reinforcement learning approach. \u003cem\u003eSci Rep\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 2713 (2021).\u003c/li\u003e\n\u003cli\u003eJ. R. Donoso, \u003cem\u003eet al.\u003c/em\u003e, Emergence of complex dynamics of choice due to repeated exposures to extinction learning. \u003cem\u003eAnim Cogn\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, 1279\u0026ndash;1297 (2021).\u003c/li\u003e\n\u003cli\u003eJ. M. Peschken, L. Hahn, R. Pusch, J. Rose, Context is Learned, not Given. [Preprint] (2025). Available at: https://www.researchsquare.com/article/rs-5682968/v1 [Accessed 3 March 2025].\u003c/li\u003e\n\u003cli\u003eJ. J. Medina Peschken, \u0026ldquo;Context as a learned property of stimuli in extinction learning\u0026rdquo; (Animal Study Registry, 2023).\u003c/li\u003e\n\u003cli\u003eT. Otto, J. Rose, The open toolbox for behavioral research. \u003cem\u003eBehav Res\u003c/em\u003e \u003cstrong\u003e56\u003c/strong\u003e, 4522\u0026ndash;4529 (2023).\u003c/li\u003e\n\u003cli\u003eD. H. Brainard, The Psychophysics Toolbox. \u003cem\u003eSpatial Vis\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 433\u0026ndash;436 (1997).\u003c/li\u003e\n\u003cli\u003eJ. R. Donoso, \u003cem\u003eet al.\u003c/em\u003e, Emergence of complex dynamics of choice due to repeated exposures to extinction learning. \u003cem\u003eAnim Cogn\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, 1279\u0026ndash;1297 (2021).\u003c/li\u003e\n\u003cli\u003eJ. Packheiser, O. G\u0026uuml;nt\u0026uuml;rk\u0026uuml;n, R. Pusch, Renewal of extinguished behavior in pigeons (Columba livia) does not require memory consolidation of acquisition or extinction in a free-operant appetitive conditioning paradigm. \u003cem\u003eBehavioural Brain Research\u003c/em\u003e \u003cstrong\u003e370\u003c/strong\u003e, 111947 (2019).\u003c/li\u003e\n\u003cli\u003eJ. Packheiser, J. R. Donoso, S. Cheng, O. G\u0026uuml;nt\u0026uuml;rk\u0026uuml;n, R. Pusch, Trial-by-trial dynamics of reward prediction error-associated signals during extinction learning and renewal. \u003cem\u003eProgress in Neurobiology\u003c/em\u003e \u003cstrong\u003e197\u003c/strong\u003e, 101901 (2021).\u003c/li\u003e\n\u003cli\u003eM. E. Bouton, Context and behavioral processes in extinction. \u003cem\u003eLearn Mem\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 485\u0026ndash;494 (2004).\u003c/li\u003e\n\u003cli\u003eB. L. Thomas, N. Larsen, J. J. B. Ayres, Role of context similarity in ABA, ABC, and AAB renewal paradigms: Implications for theories of renewal and for treating human phobias. \u003cem\u003eLearning and Motivation\u003c/em\u003e \u003cstrong\u003e34\u003c/strong\u003e, 410\u0026ndash;436 (2003).\u003c/li\u003e\n\u003cli\u003eS. Y.-S. Khoo, J. M. Sciascia, A. Brown, N. Chaudhri, Comparing ABA, AAB, and ABC Renewal of Appetitive Pavlovian Conditioned Responding in Alcohol- and Sucrose-Trained Male Rats. \u003cem\u003eFront. Behav. Neurosci.\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 5 (2020).\u003c/li\u003e\n\u003cli\u003eW. Clark, M. Colombo, Seeing the Forest for the Trees, and the Ground Below My Beak: Global and Local Processing in the Pigeon\u0026rsquo;s Visual System. \u003cem\u003eFront. Psychol.\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 888528 (2022).\u003c/li\u003e\n\u003cli\u003eF. A. Soto, E. A. Wasserman, Mechanisms of object recognition: what we have learned from pigeons. \u003cem\u003eFront. Neural Circuits\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e (2014).\u003c/li\u003e\n\u003cli\u003e\u0026ldquo;The Avian Visual System: Overview\u0026rdquo; in \u003cem\u003eHow Animals See the WorldComparative Behavior, Biology, and Evolution of Vision\u003c/em\u003e, (Oxford University Press, 2012), pp. 473\u0026ndash;482.\u003c/li\u003e\n\u003cli\u003eM. A. McDevitt, E. Fantino, Visual stimulus compounding with pigeons. \u003cem\u003eBehavioural Processes\u003c/em\u003e \u003cstrong\u003e38\u003c/strong\u003e, 265\u0026ndash;275 (1996).\u003c/li\u003e\n\u003cli\u003eK. Goto, S. Watanabe, The whole is equal to the sum of its parts: Pigeons (Columba livia) and crows (Corvus macrorhynchos) do not perceive emergent configurations. \u003cem\u003eLearn Behav\u003c/em\u003e \u003cstrong\u003e48\u003c/strong\u003e, 53\u0026ndash;65 (2020).\u003c/li\u003e\n\u003cli\u003eD. I. Brooks, R. G. Cook, K. Goto, Perceptual grouping and detection of trial-unique emergent structures by pigeons. \u003cem\u003eAnim Cogn\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, 717\u0026ndash;729 (2022).\u003c/li\u003e\n\u003cli\u003eK. K. Cavoto, R. G. Cook, Cognitive precedence for local information in hierarchical stimulus processing by pigeons. \u003cem\u003eJournal of Experimental Psychology: Animal Behavior Processes\u003c/em\u003e \u003cstrong\u003e27\u003c/strong\u003e, 3\u0026ndash;16 (2001).\u003c/li\u003e\n\u003cli\u003eJ. Emmerton, J. C. Renner, Local rather than global processing of visual arrays in numerosity discrimination by pigeons (Columba livia). \u003cem\u003eAnim Cogn\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 511\u0026ndash;526 (2009).\u003c/li\u003e\n\u003cli\u003eU. Aust, E. Braun\u0026ouml;der, Transfer between local and global processing levels by pigeons (Columba livia) and humans (Homo sapiens) in exemplar- and rule-based categorization tasks. \u003cem\u003eJournal of Comparative Psychology\u003c/em\u003e \u003cstrong\u003e129\u003c/strong\u003e, 1\u0026ndash;16 (2015).\u003c/li\u003e\n\u003cli\u003eK. Goto, A. J. Wills, S. E. G. Lea, Global-feature classification can be acquired more rapidly than local-feature classification in both humans and pigeons. \u003cem\u003eAnimal Cognition\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, 109\u0026ndash;113 (2004).\u003c/li\u003e\n\u003cli\u003eZ. Rezvani, A. Katanforoush, H. Pouretemad, Global precedence changes by environment: A systematic review and meta-analysis on effect of perceptual field variables on global-local visual processing. \u003cem\u003eAtten Percept Psychophys\u003c/em\u003e \u003cstrong\u003e82\u003c/strong\u003e, 2348\u0026ndash;2359 (2020).\u003c/li\u003e\n\u003cli\u003eS. P. Le\u0026oacute;n, A. M. G\u0026aacute;mez, J. M. Rosas, Mechanisms of Contextual Control when Contexts are Informative to Solve the Task. \u003cem\u003eSpan. j. psychol.\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 10\u0026ndash;19 (2012).\u003c/li\u003e\n\u003cli\u003eS. Lucke, H. Lachnit, S. Koenig, M. Uengoer, The informational value of contexts affects context-dependent learning. \u003cem\u003eLearn Behav\u003c/em\u003e \u003cstrong\u003e41\u003c/strong\u003e, 285\u0026ndash;297 (2013).\u003c/li\u003e\n\u003cli\u003eS. Lucke, H. Lachnit, M. C. St\u0026uuml;ttgen, M. Uengoer, The impact of context relevance during extinction learning. \u003cem\u003eLearn Behav\u003c/em\u003e \u003cstrong\u003e42\u003c/strong\u003e, 256\u0026ndash;269 (2014).\u003c/li\u003e\n\u003cli\u003eN. J. Mackintosh, An Analysis of Overshadowing and Blocking. \u003cem\u003eQuarterly Journal of Experimental Psychology\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, 118\u0026ndash;125 (1971).\u003c/li\u003e\n\u003cli\u003eN. J. Mackintosh, Overshadowing and stimulus intensity. \u003cem\u003eAnimal Learning \u0026amp; Behavior\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, 186\u0026ndash;192 (1976).\u003c/li\u003e\n\u003cli\u003eC. Bonardi, E. Mondrag\u0026oacute;n, B. Brilot, D. J. Jennings, Overshadowing by fixed- and variable-duration stimuli. \u003cem\u003eQuarterly Journal of Experimental Psychology\u003c/em\u003e \u003cstrong\u003e68\u003c/strong\u003e, 523\u0026ndash;542 (2015).\u003c/li\u003e\n\u003cli\u003eD. J. Jennings, C. Bonardi, K. Kirkpatrick, Overshadowing and stimulus duration. \u003cem\u003eJournal of Experimental Psychology: Animal Behavior Processes\u003c/em\u003e \u003cstrong\u003e33\u003c/strong\u003e, 464\u0026ndash;475 (2007).\u003c/li\u003e\n\u003cli\u003eK. J. Leising, D. Garlick, A. P. Blaisdell, Overshadowing between landmarks on the touchscreen and in arena with pigeons. \u003cem\u003eJournal of Experimental Psychology: Animal Behavior Processes\u003c/em\u003e \u003cstrong\u003e37\u003c/strong\u003e, 488\u0026ndash;494 (2011).\u003c/li\u003e\n\u003cli\u003eM. R. Horne, J. M. Pearce, Potentiation and overshadowing between landmarks and environmental geometric cues. \u003cem\u003eLearn Behav\u003c/em\u003e \u003cstrong\u003e39\u003c/strong\u003e, 371\u0026ndash;382 (2011).\u003c/li\u003e\n\u003cli\u003eM. L. Spetch, Overshadowing in landmark learning: Touch-screen studies with pigeons and humans. \u003cem\u003eJournal of Experimental Psychology: Animal Behavior Processes\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 166\u0026ndash;181 (1995).\u003c/li\u003e\n\u003cli\u003eR. Deery, S. Commins, Landmark Distance Impacts the Overshadowing Effect in Spatial Learning Using a Virtual Water Maze Task with Healthy Adults. \u003cem\u003eBrain Sciences\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 1287 (2023).\u003c/li\u003e\n\u003cli\u003eJ. M. Rosas, J. E. Callejas Aguilera, M. M. Ramos \u0026Aacute;lvarez, M. J. Fern\u0026aacute;ndez Abad, Revision of retrieval theory of forgetting: What does make information context-specific? \u003cem\u003eInternational Journal of Psychology and Psychological Therapy\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, 147\u0026ndash;166 (2006).\u003c/li\u003e\n\u003cli\u003eJ. M. Rosas, J. E. Callejas-Aguilera, Context switch effects on acquisition and extinction in human predictive learning. \u003cem\u003eJournal of Experimental Psychology: Learning, Memory, and Cognition\u003c/em\u003e \u003cstrong\u003e32\u003c/strong\u003e, 461\u0026ndash;474 (2006).\u003c/li\u003e\n\u003cli\u003eP. M. Ogallar, M. M. R. \u0026Aacute;lvarez, J. A. Alcal\u0026aacute;, M. M. M. Fern\u0026aacute;ndez, J. M. Rosas, Attentional perspectives on context-dependence of information retrieval. \u003cem\u003eInternational Journal of Psychology and Psychological Therapy\u003c/em\u003e \u003cstrong\u003e17\u003c/strong\u003e, 121\u0026ndash;136 (2017).\u003c/li\u003e\n\u003cli\u003eE. M. O\u0026rsquo;Donoghue, L. Castro, E. A. Wasserman, Hierarchical and configural control in conditional discrimination learning. \u003cem\u003eJournal of Experimental Psychology: Animal Learning and Cognition\u003c/em\u003e \u003cstrong\u003e48\u003c/strong\u003e, 370\u0026ndash;382 (2022).\u003c/li\u003e\n\u003cli\u003eP. M. Blough, Attentional priming and visual search in pigeons. \u003cem\u003eJournal of Experimental Psychology: Animal Behavior Processes\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 358\u0026ndash;365 (1989).\u003c/li\u003e\n\u003cli\u003eS. L. Gray, M. A. J. Qadri, D. I. Brooks, R. G. Cook, Use of different attentional strategies by pigeons and humans in multidimensional visual search. \u003cem\u003eJournal of Experimental Psychology: Animal Learning and Cognition\u003c/em\u003e \u003cstrong\u003e48\u003c/strong\u003e, 46\u0026ndash;59 (2022).\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[{"identity":"2c6fe9e2-a479-4b7b-a1a3-890f57d6757f","identifier":"10.13039/501100001659","name":"Deutsche Forschungsgemeinschaft","awardNumber":"DFG: 316803389","order_by":0}],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Ruhr University Bochum","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Extinction learning, ABA renewal, Context, Pigeons, Operant conditioning, Stimuli competition","lastPublishedDoi":"10.21203/rs.3.rs-6146552/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6146552/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eIn extinction learning, contextual renewal occurs when an extinguished response reemerges due to a contextual shift. But what determines which stimuli function as context? Rather than being an inherent property of the environment, we propose that context emerges from competition among multiple stimuli. Using a novel paradigm in pigeons, we systematically assessed the relative influence of local, spatial, and global stimuli in driving contextual renewal. Our results reveal that renewal depends on the competitive dynamics between these stimuli, rather than any single cue acting as context by default. This challenges traditional definitions of context as a passive background and suggests that context should be operationalized based on its interaction with other available cues. Understanding these competitive mechanisms provides a framework for studying how attention and learning shape contextual control in extinction.\u003c/p\u003e","manuscriptTitle":"Context Emergence through Stimulus Competition in Extinction Learning","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-03-19 05:02:47","doi":"10.21203/rs.3.rs-6146552/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"76d1f312-2c4a-4d68-ba71-7ac096b9cc94","owner":[],"postedDate":"March 19th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":45122144,"name":"Psychology"}],"tags":[],"updatedAt":"2025-03-19T05:02:48+00:00","versionOfRecord":[],"versionCreatedAt":"2025-03-19 05:02:47","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6146552","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6146552","identity":"rs-6146552","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}