Modelling of what-where-when episodic-like memories in rats

Episodic memories contain information about the nature of an event, the place where it happened and the time when it occurred. In animals, the term “episodic-like memory” is preferred to refer to mnemonic instances containing these three features, commonly referred to as “what-where-when”. Models to study episodic-like memory have been proposed in corvidae and rodents, although their use in the neuroscience research has been limited due to certain limitations and potential ambiguities. While the neurological correlates of “what-where-when” have been identified in neuronal types such as place and time cells, it is unclear how they contribute to form a unitary representation, or how this information can be accessed during memory recall, either holistically or differentially. Here, we outline two new behavioural paradigms based on the everyday memory task that we have developed to model what and when components as well as ‘where’ information. In experiment 1 (E1), we demonstrate that rats are able to learn two distinct food positions on a daily basis and retrieve them independently. In E2, we establish that rats can learn that two flavours are replenished at different times after an initial sampling, thus use the temporal component to guide their decision making. These two tasks can therefore provide the basis to study how the item, location and time information of a memory are stored and accessed by the brain. This should be observable in single-unit recording or calcium-imaging studies. Kayleigh Kanakis 1, Richard GM Morris 2, Francesco Gobbo 2,3* 1 School of Psychology and Neuroscience, College of Medical, Veterinary, and Life Sciences, University of Glasgow, G12 8QQ, Glasgow (UK) 2 Centre for Discovery Brain Sciences, the University of Edinburgh, 1 George Square, EH8 9JZ, Edinburgh (UK) 3 UK Dementia Research Institute, the University of Edinburgh, 1 George Square, EH8 9JZ, Edinburgh (UK) Correspondence: [email protected] not-yet-known not-yet-known not-yet-known unknown 1 Abstract Episodic memories contain information about the nature of an event, the place where it happened and the time when it occurred. In animals, the term “episodic-like memory” is preferred to refer to mnemonic instances containing these three features, commonly referred to as “what-where-when”. Models to study episodic-like memory have been proposed in corvidae and rodents, although their use in the neuroscience research has been limited due to certain limitations and potential ambiguities. While the neurological correlates of “what-where-when” have been identified in neuronal types such as place and time cells, it is unclear how they contribute to form a unitary representation, or how this information can be accessed during memory recall, either holistically or differentially. Here, we outline two new behavioural paradigms based on the everyday memory task that we have developed to model what and when components as well as ‘where’ information. In experiment 1 (E1), we demonstrate that rats are able to learn two distinct food positions on a daily basis and retrieve them independently. In E2, we establish that rats can learn that two flavours are replenished at different times after an initial sampling, thus use the temporal component to guide their decision making. These two tasks can therefore provide the basis to study how the item, location and time information of a memory are stored and accessed by the brain. This should be observable in single-unit recording or calcium-imaging studies. 1 Abstract 1 Introduction Among the different forms of memory, episodic memory is perhaps the closest to our everyday concept of remembering (Squire, 1984). Memory of past experiences contain information about what happened, where and when it happened, plus a number of pieces of information such as context and temporal sequences (Nyberg et al., 1996). When it comes to human experience, resurrecting such mental representations often described as “mental time travel” (Tulving, 1985), reflecting personal involvement. This concept has been extended to “episodic-like memory” to include its behavioural manifestation in non-human animals, as it is debated and fundamentally unknown, whether animal experience can involve an analogous sense of self (Tulving, 2005). Awareness may not be even necessary to remember events, but episodic-like facets of memory have been demonstrated in primates (Gaffan, 1974), birds (Clayton & Dickinson, 1998), rodents (Babb & Crystal, 2006a), and even invertebrates (Jozet-Alves et al., 2013). Everyday memories, at least in humans, rely on the integrity of the hippocampus, as suggested by studies in amnesic patients such as H.M. (Scoville & Milner, 1957). Although scene reconstruction may be part of the reason why hippocampal integrity is required (Hassabis & Maguire, 2007), it is generally accepted that hippocampal function is associated with episodic memory, and, by extension, episodic-like memory (Allen & Fortin, 2013). In rodents, episodic-like memory is sometimes modelled with novel object recognition paradigms and variants thereof. However, the simplest such protocols can involve but are not dependent on the hippocampus (Ennaceur & Delacour, 1988; Langston & Wood, 2010). Tasks of higher complexity, such as object-in-place, are hippocampal-dependent (Aggleton & Nelson, 2020). Recency protocols using navigation in watermazes and arenas are also considered episodic-like in nature, and sensitive to hippocampal inactivation (Steele & Morris, 1999; Bast et al., 2005). Babb & Crystal (2006a) have used food rewards in particular positions of a radial arm maze to explicitly model what-where associations, and explored their sensitivity to devaluation. The same setup has also been extended to what-where-when associations (Roberts et al., 2008; Zhou & Crystal, 2009) using a depletion-replenishment paradigm, also used by Clayton et al. (2001) in scrub-jays, which currently constitutes the benchmark for episodic-like tests in animals. Unlike Clayton & Dickinson (1998), these tasks lack a clear separation between (a) decision making in advance and (b) on-line execution of the trial. This may make them unsuitable to study the separable neural representations associated with planning and remembering. Several studies have provided evidence of task-related activity, as well as replay of past, correct, and alternative directions in navigation tasks (Ainge et al., 2007; Gillespie et al., 2021; Gobbo et al., 2022). The event arena paradigm, where animals learn to retrieve food in a specific location in a two-dimensional arena, offers such a possibility, as animals enter the typically large arena from external startboxes (Bast et al., 2005; Gobbo et al., 2022). Recently, it has been used to model different navigation strategies (Broadbent et al., 2020) and contextual information (Prodan et al., 2022). Recordings from the hippocampus and associated areas have found evidence for possible neural correlates of the various type of information, such as place cells (O’Keefe, 1976), grid cells (Fyhn et al., 2004), item-position cells (Komorowski et al., 2009), and time cells (Eichenbaum, 2014). To what extent non-spatial information is represented in the hippocampus, and how these different aspects interact with each other is still largely unknown (McKenzie et al., 2016; O’Keefe & Krupic, 2021), in part due to the lack of suitable behavioural paradigms. Here, we sought to close this gap by employing a modified protocol in the everyday arena to model episodic-like what-where-when information. We based our paradigm on an allocentric definition of coordinates, and used different flavours of food to model episode content (Day et al., 2003; Tse et al., 2007). 2 Materials and Methods 2.1 Animals Twelve young male Lister-Hooded rats were used in this experiment. They were purchased from Charles River Laboratories © (Currie, UK) and housed two or four per cage. At the start of the experiment the rats were 2.5 months old and had a mean weight of 285.67 g. Rats were maintained in a 12-hour light/12-hour dark cycle. For the first week, water and food was available ad libitum . The animals were then food restricted and maintained at 85-90% of their ad-libitum weight (being fed around 22 g of food per day per animal). Experimental procedures in this study were performed strictly in line with the Animals (Scientific Procedures) Act 1986 in compliance with British Law and regulations and also the European Communities Council Directive of 24 November 1986 (86/609/EEC) legislation. 2.2 Apparatus and materials The everyday apparatus is a 1.5×1.5 m arena, comprising a 7×7 grid of square tiles of 20 cm per side on a 5 cm frame. It has walls made of transparent plexiglass. Rats enter this arena from any one of three different start boxes - south (S), east (E), and west (W). An identical but differentially placed box serves as the home-box (North). Six cylindrical plexiglass sandwells (6 cm diameter x 6 cm depth, Adam Plastics Ltd.) with a spherical hollow insert (6 cm diameter x 4 cm depth) were used to contain flavoured food as described in Tse et al. (2007). Supreme Mini-Treats 1 gm flavoured food pellets purchased from Bio-Serv (item codes are from LBS Serving Biotechnology, UK) were used: banana (1024041), very berry (1024045), chocolate (1024043), piña colada (1024044), and marshmallow (1024043). In the experiments, these treats were cut in half to produce 0.5 g pellets. The rewarded plexiglass sandwell was the only one with accessible food. To maintain uniformity of their smell, the others contained a mixture of the flavours that were used in each experiment (five pellets of each flavour), but these were located in an inaccessible compartment below the hollow compartment, albeit with accessible smell through small holes. Non-accessible pellets were refreshed routinely to ensure they maintained a constant smell. Sandwells were covered in sand (‘Cage proud’ bird sand, Pettex Ltd). Two intra-maze objects (a black-painted 40-cm water bottle and a 48-cm stack of glued golf balls), were present in the arena, and several 2D and 3D extra-maze objects were present in the room. Extramaze cues were about 40- to 60-cm in size. Sessions were recorded using a ceiling mounted camera (CCTVFirst), positioned centrally above the arena floor, and OBS recording software and DeckLink Mini Recorder cards (Blackmagic Design Pty, Ltd). The experimenter room and control panel were separated from the arena room by blackout curtains. A 70% ethyl alcohol solution was used to clean the arena between each animal. 2.3 Habituation To familiarise the animals with food, 10 g of flavoured pellets were left in the rats’ cages overnight but buried in sand to encourage digging. Dietary restriction was put in place to maintain animals at 85-90% of free-food weight based on a previously established weight curve (Gobbo et al., 2022). During habituation, the rats were individually placed in the open-field arena to be familiarised with the environment and experimenter handling. Lighting was dimmed to encourage exploration. Animals were encouraged to leave the start-box, find the pellets, and return to the home-box to eat. 2.4 Flavour preference test Six rats chosen randomly were used to evaluate their food preference. Two sandwells were open at the same time with the flavours of banana (B), very berry (Vb), or chocolate (C) all accessible. Three trials were run per day for three consecutive days. Two sandwells containing pellets of two different flavours were located equidistant from the start-box (S). Rats were placed in the start-box, allowed to enter the arena, approach one or other sandwell with accessible food, and choose one of two flavours. Once the subject chose a flavour, they retrieved the food pellet and then entered the N box to eat it, followed by re-entering the arena to choose a second pellet, either of the either same or a different flavour. This was repeated twice in each session until the food was depleted or 10 min of time had elapsed. A Food Preference (FP) score was calculated as follows: the two flavours were assigned +1 or -1 values, and for each trial the corresponding value was assigned based on the flavour of the first pellet chosen by the rat. The FP score is the averaged score across the total number of test sessions. 2.5 Experiment 1: What-Where (Flavours changing daily) The behavioural protocol was similar to Broadbent et al. (2020), with modifications to include two flavours on a daily basis. In each session, the animals learned a daily novel location for each of the two flavours. First, subjects completed a sample phase corresponding to three trials per flavour (Figure 1a). Each sample trial began from one of the three startboxes (W, S, E) which were randomly assigned. The animal navigated to the home box (N) after retrieving the sample pellet at the sandwell. It then had the opportunity to retrieve a second pellet in the arena, and return to N. At the beginning of each trial, the animal received a cue pellet in the startbox, of the same flavour present in the correct sandwell to instruct it what flavour to expect. In each sample trial (STs), three accessible reward pellets were placed in the bottom of the sand-filled spherical bowl; the remaining sandwells did not contain sand. Trials with flavour 1 and flavour 2 were alternated for a total of six STs per session (three for each flavour); the first flavour was selected randomly. After a time delay of 50-75 mins, the choice phase followed. In the choice phase of each session, one of the two flavours was selected randomly, and the subject’s memory was tested of whether it could recall the correct location of the flavour cued in the start-box. Two trials were run in the choice phase. The choice of flavour to be identified in choice trials (CTs) for each session was determined randomly, with the condition that, in each session, half of the animals were tested on one flavour and half on the other. In CTs, all wells appeared identical and were filled with sand. The usage of the startboxes in STs and CTs, the location of the rewarded sandwells, the identity of the first flavour in the sample phase, and the flavour tested in CTs were randomised and counterbalanced between animals using a random generator (https://commentpicker.com/random-number-generator.php); the combinations are reported in the deposited data. All sandwells contained a mixture of excess pellets in an inaccessible compartment (Figure 1b) to ensure that they all smelled the same to the animals. The rats were randomly assigned to two groups and trained on alternating days for an initial total of 15 sessions each. Group 1 was trained with chocolate and banana flavours, while group 2 two had very-berry and banana. Then, on session 16, we then used two novel flavours to which the animals had never previously been exposed: very-berry (Vb) and marshmallow (MM) for group 1, and chocolate (C) and piña colada (PC) for group 2. Session 16 was run in the same way as earlier, except now with these novel flavours. 2.6 Experiment 2: What-Where-When Pretraining and food preference (S1-S5). After completing Experiment 1, the aim was to establish the role of “time” in memory. Procedurally, the animals were given a two-week break, whilst placed on free food. At the beginning of Experiment 2 (E2), they were 6 months old with a mean weight of 479.58 g. In E2 the same arena was used, but the positions of the sandwells (map) was different. Six distinct maps were used for the 12 animals. The sample phase from the above protocol was modified as follows: two flavours (2 𝑥 0.5 g pellets each) from three possible combinations (Vb/MM, C/MM, or C/Vb) were accessible at the same time, and all wells were sand filled. The positions of the two flavours remained fixed for a given animal across sessions. The start-box position changed within and between sessions. The subject completed two STs where it located both sets of flavours until the STs were depleted. Over 5-days, the subject was allowed to form flavour-place paired associations. Tiles were randomly replaced between sessions and the arena wiped down with 70% ethanol to help reduce potential olfactory confounds. Training sessions (S6-31). From S6, sessions comprised one sample phase, followed by two CTs. One CT was after a short (10 min; CT 10 min) and the other after a long (3 hours; CT 3h) retention interval (time delay between exposure and recall). Only one of the two flavours was replenished and available in the CT after 10 min; the other was replenished after 3 hrs. In the CT at 3h, the previously unavailable flavour was replenished, and the other flavour depleted. The identity of the flavours was randomly assigned to each animal. This identity was consistent for the length of Experiment 2, e.g. for a given rat MM was always replenished at 10 min, and Vb at 3 hrs. Two pellets (0.5 g/pellet) of the available flavour were placed at the bottom of the sandwell in the accessible compartment. Once the first pellet was found, the rat carried it to the home-box to eat, and then was allowed to re-enter the arena to obtain the second. Trials ended once both pellets had been located and the rat returned to the home-box. Sessions were run for a total of 31 days (n=12). The last two sessions were run as before, but the CT after 10 min was omitted. Probe trial sessions (S22, 24, 25). To confirm that animals were not using olfactory cues in making choices, probe trial sessions (PT) were conducted during sessions 22, 24, and 25, as described in Tse et al. (2007), in lieu of the CT at 3h. In probe trials, all six sandwells were non-rewarded and contained only inaccessible pellets. The key issue was where the animal focused his digging. Animals were tested on their memory for 2 min., and at the end of the trial the north door was opened and, once the animal had entered it, the rat returned to the home-box. To prevent extinction, the sandwell that would normally contain the reward was replenished at the bottom with two pellets at the end of the PT, but a regular CT was performed. Digging time refers to the time during which the animals’ forepaws moved the sand with intention to locate pellets beneath. Time spent sniffing, nose-poking, and walking over the sandwell was not considered ‘digging’. 2.7 Histology One-hour after the CT 3h, each rat was deeply anaesthetised with pentobarbital and perfused with cold phosphate buffered saline (PBS, Merck 524650) and 4% formaldehyde; the brain was fixed overnight in formaldehyde, then immersed in 30% sucrose PBS for 48 hrs at 4℃. 60-μm-thick coronal sections were cut with a cryostat (ca. Bregma –3.60 mm). Sections were incubated in 10% normal donkey serum (NDS, Merck 566460), 0.3% Triton X-100 PBS for 60’, then overnight with rabbit anti-c-Fos IgG (1:1000 dilution, Synaptic Systems 226308) in PBS with 10% NDS and 0.1% Triton X-100. Three 5’ washes with PBS were completed before applying the secondary antibody, donkey anti-guinea pig Cy3 conjugate (1:200 dilution, Merck AP193C), for 120 mins. Slices were washed as before in PBS for 5 mins each. The slices were slide-mounted with Fluoroshield with DAPI (Abcam ab104135) mounting medium. 5-μm z-stack images were captured on a Nikon Eclipse Ti inverted confocal microscope with 20X objective. After projecting the z-stack, the same threshold was set to all images and neurons with positive signals were quantified as c-Fos+/DAPI+ cells with ImageJ CellCounter Plugin (NIH). 2.8 Data Analysis Data were analysed during the experiment and rescored blind by two independent users. Data was analysed with GraphPad Prism (v9.3.1). The Performance Index (PI) was calculated using the number of sandwells at which the rats dug before locating the correct sandwell, calculated as 100*(5-errors)/5, where chance performance is 50%. Latency is the time taken from exiting the start-box to the start of digging at the correct sandwell (sec). Accuracy was assigned a value of 1 if rats dug at either of the previously rewarded sandwells from the ST in the CTs, and 0 if they dug at any other. Conversion to a score was calculated as the summed total of each animal per session over the mean (n=12). In probe sessions, the time spent digging at the correct (or previously rewarded sandwell) is expressed as the fraction of the total digging time at all sandwells (Fraction dig time). We also expressed the total dig time at the correct sandwell (3 h sandwell) over the total time spent digging at the correct or alternative (10 min flavour) sandwell (Discrimination dig time). Data were analysed using one-way analysis of variance (ANOVA), mixed linear models, or paired sample t-tests, as appropriate. Performance and accuracy across sessions were assessed by repeated measures two-way ANOVA with Greenhouse-Geisser corrections and by one-way ANOVA with Wilcoxon rank-sum test with a Bonferroni correction as appropriate. For the immunohistochemical analysis, the mean number of stained nuclei from two sections in the region of interest (dorsal CA1 and DG) was used for group comparisons of c-Fos expression. Results were analysed using a one sample t-test. Statistical significance was set to α=0.05. 2.9 Data Availability Data from this publication can be accessed from the University of Edinburgh DATASTORE. Original videos, images and records are available upon request from the corresponding author. 3 Results We sought to determine if the rats were able to form and retrieve independent episodic memories containing where-what information (in E1) and where-what-when (in E2). Our experimental setup is based on the everyday arena paradigm (Bast et al., 2005), where the animals learn to retrieve food pellets from one of multiple sandwell positions in a two dimensional arena. We modelled the what aspect of memory with food pellets of different flavours as used in Day et al. ( 2003) and Tse et al. (2007). In the everyday memory task, there is long term-memory of the apparatus, and the environment, as well the task outline (Figure 1A). Which sandwell is rewarded each day, instead, constitutes a critical aspect of the episodic memory. The intensity and the perceived importance of these memories determines how long they last; this was observed to range from hours to days (Wang et al., 2010). 3.1 Food preference We first determined if rats had any preference towards one of the flavours, which could otherwise bias the search towards the preferred flavour in the following experiments. Rats were placed in a box and were allowed access to the arena (see Methods section 2.2). Two sandwells were open, each containing three pellets of the same flavour, but the flavours differed between the sandwells. Animals were left to choose which flavour to consume and in which order. Over three trials on different days, there was no significant difference with respect to which flavour was picked first (Flavour Preference Ratio, Figure 1C) (Welch’s ANOVA F(3,18)=1.489 P=0.27), or in the amount eaten (Brown-Forsythe ANOVA F(3,18)=1.226 P=0.34) of the two flavours (Supplementary Figure S1a). We therefore concluded that the flavours used were equally appealing to the rats and were used in the following experiments. 3.2 Experiment 1: What-Where-When – focusing on the “where” aspect. In the everyday memory task, animals learned to retrieve food pellets from one of six possible sandwells. The location of the reward changes every day, constituting one aspect of the episodic (or recency) task (Bast et al., 2005; Wang et al., 2010). Because positions change from day to day, the animals need to remember where they last retrieved a certain flavour that day, making this a recency task. To promote the use of an allocentric map, they entered the arena from one of three possible Start Boxes, in which they received a cue pellet. After retrieving the food pellet from the correct sandwell, they carried it to a Goal Box (conventionally located on the North side of the arena) (Broadbent et al., 2020). In Experiment 1, we tested the possibility that rats could form and retrieve two independent episodic memories represented by the positions of food pellets of two different flavours. In the STs, they were given the opportunity to learn the flavour-sandwell associations after being cued with a pellet of the corresponding flavour in the startbox. After 90 min., a CT was performed, where their memory was tested: a cue pellet of the test flavour was given to them in the Start Box, and the animals had to choose the correct sandwell for that flavour among six identical sandwells (Figure 1a). Two groups of animals were trained with different combinations of flavours (either Chocolate-Banana or Banana-Very-Berry). To ensure that all sandwells had a similar olfactory profile, they all contained a mixture of the two flavours in the bottom, inaccessible compartment (Figure 1b). Rats in the two groups exhibited similar learning curves (repeated measures REML linear model: Session factor F(df 4.91, 56.25)=2.435 P=0.047); Flavour factor F(df 1,149)=0.11 P=0.74. Their performance was initially at chance, but it reached above 80% in the last group of sessions (S12-15) (repeated measures 2-way ANOVA F(8.480, 197.9)=3.357, P= 0.001, Figure 1d and Supplementary Figure S1b). Similarly, latency showed a decreasing trend across sessions (test for linear trend F(1,164)=9.395 P=0.0025) (Supplementary Figure S1c,d). Accuracy increased across sessions, and was significantly different from chance during sessions 11-15 (Mann Whitney test P<0.05, corrected for multiple measures with the BKY Two-stage step-up FDR method 0.05). This rules out that the animals simply alternated between the two rewarded wells (Supplementary Figure S1e,f). One possibility is that the increased performance of the animals is in part due to familiarity towards the flavours used. To exclude this possibility, on Session 16, animals were trained with two novel flavours they had never encountered before (Very Berry-Marshmallow and Chocolate-Piña Colada, respectively). The performance with the new flavours was analogous to the performance in the previous three sessions with the old flavours (Student’s t-test t=1.077, df=10, P=0.3068, Figure 1e). Accuracy was above chance even for the new combination of flavours (P=0.0013, Mann Whitney test corrected for multiple comparisons). This confirms the episodic nature of the memory, as new information regarding the position and the flavour can be learned and retrieved independently. 3.3 Experiment 2: What-Where-When - focusing on the “When” aspect. In Experiment 2, we sought to model the time component more explicitly as in Clayton & Dickinson (1998). To do this, we designed an experiment in order that, after the rats had experienced two flavours in the STs, one of the two flavours was replenished after a short interval (10 min.), and the other after a long interval (3 hr) (Figure 2a). The main task was preceded by five habituation sessions, during which the animals familiarised themselves with two novel flavour combinations and established the “what-where” association (see Methods section). As shown in Figure 2b, rats progressively learned which flavour was replenished at 10 min, and which at 3 h. The simplest way to solve the task is to simply try one of the two rewarded locations; indeed, this was the strategy initially employed by the animals, and accuracy was not different from chance in the earlier sessions (Figure 2c). Progressively, the rats learned which of the two flavours was replenished at 10 min and which at 3 h: accuracy increased for both short-term and long-term replenishment (10 min: REML mixed model test for trend F (1, 238) = 12.31 P=0.0005; 3h: F (1, 212) = 41.92 P<0.0001) and reliably stayed significantly above chance. In sessions 28 and 29, we omitted the CT at 10 min and ran only the CT at 3 h. Animals correctly and accurately searched for the flavour replenished at 3 h (Figure 2b,c and Supplementary Figure S2b). This demonstrates that rats are not learning an order or sequence of flavours to search, but explicitly form an association between the identity of the flavour and the time of replenishment. To control for olfactory cues, in sessions 21, 24, and 25, probe trials were run where no reward was present in the sandwells in place of CT 3h. In probe trial 1 (PT1, session 21), rats dug at the two possible rewarded sandwells similarly (repeated measures one-way ANOVA F (1.699, 18.69) = 3.665, P=0.052). It is not uncommon in this type of task for animals to be confused by not finding the expected reward in the first probe trial, therefore opting to try other sandwells (Gobbo et al., 2022; Tse et al., 2023). Indeed, accuracy in PT1 was essentially at chance, and the low relative fraction time was due to continued digging at the wrong Sandwell (Supplementary Figure S2d,e). When we repeated probe trials in place of CT 3h in sessions 24 and 25 (PT2 and PT3), animals dug significantly more at the expected sandwell, i.e. the Sandwell expected to be replenished at 3 h (repeated measures one-way ANOVA PT2: F (1.180, 12.98) = 17.74 P=0.0007, PT3: F (1.191, 13.11) = 20.92 P=0.0003; Supplementary Figure S2c-e). This confirmed that rats correctly searched for the flavour rewarded at 3 h and excluded the use of olfactory cues to retrieve it. Furthermore, it demonstrates that rats correctly discriminated between the flavour replenished at 3 h and the one replenished at 10 min (Supplementary Figure S2e). 3.4 Hippocampal recruitment To confirm that the hippocampus was involved in the solution of the task in Experiment 2, we sacrificed the rats 60 min. after performing long-term retrieval (3 h Choice), as well as animals that were trained but not tested at 3 h. A significantly higher number of cells expressed cfos in the upper DG blade and CA1 pyramidal layer, that contained virtually only the cell bodies of excitatory neurons (Figure 3). This confirmed that recalling the identity of the flavour replenished at 3 h involved hippocampal activity. 4 Discussion Episodic memories have a central role in everybody’s life by shaping our experience and forming our identity. They combine information about the event itself ( what ), the location ( where ) – or sometimes the context surrounding it – and the time where it happened ( when ) (Tulving, 1985). The majority of available evidence points to the hippocampal formation and the medial temporal lobe in general as key brain areas dedicated to their processing (Nyberg et al., 1996; Nadel & Moscovitcht, 1997; McKenzie et al., 2024). They are severely affected by neurodegenerative disease such as Alzheimer’s disease (Knopman et al., 2021). In the last decades, our understanding of information processed by memory areas has advanced greatly, but we need refined models to study advanced brain functions. The theoretical discussion of whether animals possess an equivalent form of episodic memory – which has led to the definition of the concept “episodic-like” memory for noetic reasons (Tulving, 2005) – has found compelling evidence in favour of it as a concept (Clayton & Dickinson, 1998; Clayton et al., 2001; Eichenbaum et al., 2005; Babb & Crystal, 2006a). Reliable behavioural tasks to model episodic-like memories are notoriously hard to implement. Tasks such as novel object recognition, object-place memory, and object-place-context memory have - unfortunately - been found to be based on familiarity rather than explicit recall (Ennaceur et al., 1997; Wilkinson et al., 2006). More recently, episodic-like aspects such as ‘what-where’ or ‘what-when’ associations have been modelled using the framework of the E maze (Eacott et al., 2005) or the radial arm maze (Babb & Crystal, 2006a; Naqshbandi et al., 2007; Zhou & Crystal, 2009). The involvement of the hippocampus in these tasks, however, has not yet been investigated. This is an important aspect, as surprisingly many tasks can be solved in even the absence of hippocampal activity, likely due to multiple parallel memory systems (Langston & Wood, 2010; Broadbent et al., 2020; Duszkiewicz et al., 2023). The general rule seems to be that hippocampal involvement tends to decrease with the complexity of the task, or its solvability with algorithmic or stereotypical steps. The event arena is similar to the watermaze in that it is a “memory recall” task (Nonaka et al., 2017). Animals enter the arena from separate start boxes, thereby disentangling recall from the overt execution of the task following a decision (Gobbo et al., 2022). Compared to the watermaze, it has the advantage of being a dry maze, hence enabling electrophysiological or optical recordings; and is an appetitive, or positive reinforcement task, making it a preferable choice in terms of refinement. The disadvantage is that unless opportunely designed, tasks can be ambiguously solved with easier, less computationally demanding approaches that do not necessarily rely on the hippocampus (Broadbent et al., 2020). Furthermore, to date, the event arena has been used such that it only models spatial memory in an episodic way, without explicitly including temporal or item information. Here, we have built on the allocentric version of the event arena task to model more fully what-where information (Experiment 1) and what-where-when information (Experiment 2). Previous attempts to model what information with individual flavours were not conclusive as no appropriate analyses and controls were included (Day et al., 2003). In E1, we demonstrate that rats are capable of learning the location of two independent flavour-sandwell ( what-where ) combinations in a daily manner. The ability to perform above chance even with new flavours never previously encountered demonstrates that the animals rapidly form a well-rounded memory of finding a particular food in a defined location. In E2, we show that rats can learn that individual flavours in different locations are replenished at different times, mirroring experiments in scrub jays (Clayton & Dickinson, 1998). Probe trials and tests run at 3 h in the absence of 10 min tests (Figure 2) are appropriate controls for demonstrating that rats use the absolute time elapsed from the Sample Trial to decide which flavour to retrieve, rather than alternating between the two. Together, these experiments provide important paradigms for future studies to investigate how the various aspects of episodic-like memories are stored and retrieved during decision-making (Gobbo et al., 2022). Nevertheless, a number of points need to be considered. First, although both performance and accuracy rise above chance in later sessions, the learning curve is somewhat flat for the first ten sessions. In contrast, rats typically learn the single-flavoured task in about five sessions or less, when using an allocentric protocol (Broadbent et al., 2020; Gobbo et al., 2022). This raises the possibility that rats may have difficulties in understanding the two main rules necessary to identify the correct sandwell when two sandwells have to be learned together: i) that the position of the correct sandwell is the same regardless of the starting box (allocentric rule), and ii) that the flavour cued in the startbox will be present in the arena ( what rule). One possibility would be to train rats in the allocentric protocol initially with one flavour (or non-flavoured pellets) and introduce multiple flavours once the choice rule has been learned. Second, while we randomised the flavour identity in the first ST, we alternated the two flavours in samples 1-6 (see Methods). This might create an expectation in the animals for flavours to be alternated; an improvement of the protocol would be to randomise the order of flavours across the six STs. Third, E2 used the replenishment time for animals to decide for which flavour they should search. Rats were capable of performing the task, extending the results of Babb & Crystal (2006b, 2006a) and Naqshbandi et al. (2007). It should be noted that we used relative time from a pair of STs (one for each flavour, see Methods) designed to inform the animal about when the food was available, and created an expectation based on learned knowledge about when each flavour would be replenished. This is different from an absolute concept of time that is generally believed to be associated with the mental time travel of human memories. Of note, although there have been claims about the possibility of using circadian time to inform choices (Zhou & Crystal, 2009), these have not been supported by evidence from other experiments (Roberts et al., 2008). Incidentally, elapsed time might be a more useful feature to remember in practical terms and indeed time cells seem to map the time since a salient event (MacDonald et al., 2011), and the concept of when in human episodic-memory has, during time travel, itself a marked contextual component (Tulving, 1985; Prescott et al., 2019; McKenzie et al., 2024). In conclusion, the experiments described here provide new information about the possibility of explicitly integrating multiple pieces of information in mnemonic tasks based on an allocentric, hippocampus dependent paradigm (Broadbent et al., 2020). Hopefully, these will be useful in the design of experiments to advance our understanding of episodic memories and how they guide complex information processing when taking decisions based on past experiences. Figure legends Figure 1 - Experiment 1. ( a ) Schematic of a typical session in Experiment 1. ( b ) Design of the sandwells used in the study, indicating the accessible (A) and inaccessible (I) compartments. ( c ) Flavour preference for Chocolate (C)/Banana (B), Chocolate (C)/Very berry (Vb), and Very berry (Vb)/Banana (B) combinations. ( d ) Performance (inverse of the number of errors) for various sessions in Choice trials. Lines indicate the performance in the two groups of animals. Boxed session indicates the Performance in S16 where new flavours where used. ( e ) Performance in the last three sessions with old flavours and in the session with new flavours. Data (bars or empty dots) are presented as mean±sem, with filled dots representing individual animals. Figure 2 - Experiment 2. ( a ) Schematic of a typical session in Experiment 2. ( b ) Performance (inverse of the number of errors) for various sessions in the 10 min Choice trial (red) and 3 h choice trial (blue). ( c ) Accuracy for the 10 min Choice trial (red) and 3 h Choice trial (blue). Asterisks indicate sessions where the accuracy was significantly above chance according to colour code (P<0.05 Mann-Whitney U test, adjusted for multiple comparisons using the false discovery rate with the Two-stage step-up procedure of Benjamini, Krieger, and Yekutieli; FDR 1.0%). In b,c orange shading indicates sessions where probe trials were run at 3h instead of the CT, while green shading indicates sessions where CT 10min was omitted. Data are presented as mean±sem. Figure 3 - Experiment 2. ( a ) Representative images of the CA1 area from 3 h recall and home cage animals stained for DAPI (blue) and cfos (red). ( b ) Quantification of the fraction of cfos+ cells in CA1 or DG. Bars indicate mean±sem, with dots representing individual animals. *P<0.05, **P<0.01 Student’s t-test.

1. Aggleton, J.P. & Nelson, A.J.D. (2020) Distributed interactive brain circuits for object-in-place memory: A place for time? Brain and Neuroscience Advances, 4, 239821282093347.Ainge, J.A., Tamosiunaite, M., Woergoetter, F., & Dudchenko, P.A. (2007) Hippocampal CA1 place cells encode intended destination on a maze with multiple choice points. Journal of Neuroscience, 27, 9769–9779.Allen, T.A. & Fortin, N.J. (2013) The evolution of episodic memory. Proceedings of the National Academy of Sciences of the United States of America, 110, 10379–10386.Babb, S.J. & Crystal, J.D. (2006a) Episodic-like Memory in the Rat. Current Biology, 16, 1317–1321.Babb, S.J. & Crystal, J.D. (2006b) Discrimination of what, when, and where is not based on time of day. Learning & Behavior, 34, 124–130.Bast, T., Da Silva, B.M., & Morris, R.G.M. (2005) Distinct contributions of hippocampal NMDA and AMPA receptors to encoding and retrieval of one-trial place memory. Journal of Neuroscience, 25, 5845–5856.Broadbent, N., Lumeij, L.B., Corcoles, M., Ayres, A.I., Bin Ibrahim, M.Z., Masatsugu, B., Moreno, A., Carames, J., Begg, E., Strickland, L., Mazidzoglou, T., Padanyi, A., Munoz-Lopez, M., Takeuchi, T., Peters, M., Morris, R.G.M., & Tse, D. (2020) A stable home‐base promotes allocentric memory representations of episodic‐like everyday spatial memory. Eur J Neurosci, 51, 1539–1558.Clayton, N.S. & Dickinson, A. (1998) Episodic-like memory during cache recovery by scrub jays. Nature, 395, 272–274.Clayton, N.S., Yu, K.S., & Dickinson, A. (2001) Scrub jays (Aphelocoma coerulescens) form integrated memories of the multiple features of caching episodes. Journal of Experimental Psychology: Animal Behavior Processes, 27, 17–29.Day, M., Langston, R., & Morris, R.G.M. (2003) Glutamate-receptor-mediated encoding and retrieval of paired-associate learning. Nature, 424, 205–209.Duszkiewicz, A.J., Rossato, J.I., Moreno, A., Takeuchi, T., Yamasaki, M., Genzel, L., Spooner, P., Canals, S., & Morris, R.G.M. (2023) Execution of new trajectories toward a stable goal without a functional hippocampus. Hippocampus, 33, 769–786.Eacott, M.J., Easton, A., & Zinkivskay, A. (2005) Recollection in an episodic-like memory task in the rat. Learn. Mem., 12, 221–223.Eichenbaum, H. (2014) Time cells in the hippocampus: A new dimension for mapping memories. Nature Reviews Neuroscience, 15, 732–744.Eichenbaum, H., Fortin, N.J., Ergorul, C., Wright, S.P., & Agster, K.L. (2005) Episodic recollection in animals: “If it walks like a duck and quacks like a duck….” Learning and Motivation, 36, 190–207.Ennaceur, A. & Delacour, J. (1988) A new one-trial test for neurobiological studies of memory in rats. 1” Behavioral data, Behavioural Brain Research.Ennaceur, A., Neave, N., & Aggleton, J.P. (1997) Spontaneous object recognition and object location memory in rats: the effects of lesions in the cingulate cortices, the medial prefrontal cortex, the cingulum bundle and the fornix: Exp Brain Res, 113, 509–519.Fyhn, M., Molden, S., Witter, M.P., Moser, E.I., & Moser, M.-B. (2004) Spatial Representation in the Entorhinal Cortex. Science, 305, 1258–1264.Gaffan, D. (1974) Recognition impaired and association intact in the memory of monkeys after transection of the fornix, Journal of Comparative and Physiological Psychology.Gillespie, A.K., Astudillo Maya, D.A., Denovellis, E.L., Liu, D.F., Kastner, D.B., Coulter, M.E., Roumis, D.K., Eden, U.T., & Frank, L.M. (2021) Hippocampal replay reflects specific past experiences rather than a plan for subsequent choice. Neuron, 109, 3149-3163.e6.Gobbo, F., Mitchell-Heggs, R., Tse, D., Al Omrani, M., Spooner, P.A., Schultz, S.R., & Morris, R.G.M. (2022) Neuronal signature of spatial decision-making during navigation by freely moving rats by using calcium imaging. Proc. Natl. Acad. Sci. U.S.A., 119, e2212152119.Hassabis, D. & Maguire, E.A. (2007) Deconstructing episodic memory with construction. Trends in Cognitive Sciences, 11, 299–306.Jozet-Alves, C., Bertin, M., & Clayton, N.S. (2013) Evidence of episodic-like memory in cuttlefish. Current Biology, 23 .Knopman, D.S., Amieva, H., Petersen, R.C., Chételat, G., Holtzman, D.M., Hyman, B.T., Nixon, R.A., & Jones, D.T. (2021) Alzheimer disease. Nat Rev Dis Primers, 7, 33.Komorowski, R.W., Manns, J.R., & Eichenbaum, H. (2009) Robust conjunctive item - Place coding by hippocampal neurons parallels learning what happens where. Journal of Neuroscience, 29, 9918–9929.Langston, R.F. & Wood, E.R. (2010) Associative recognition and the hippocampus: Differential effects of hippocampal lesions on object-place, object-context and object-place-context memory. Hippocampus, 20, 1139–1153.MacDonald, C.J., Lepage, K.Q., Eden, U.T., & Eichenbaum, H. (2011) Hippocampal “Time Cells” Bridge the Gap in Memory for Discontiguous Events. Neuron, 71, 737–749.McKenzie, S., Gobbo, F., & Morris, R.G.M. (2024) Theories of Hippocampal Function. In The Hippocampus Book, 2nd edn. Oxford University Press, pp. 649–722.McKenzie, S., Keene, C.S., Farovik, A., Bladon, J., Place, R., Komorowski, R., & Eichenbaum, H. (2016) Representation of memories in the cortical–hippocampal system: Results from the application of population similarity analyses. Neurobiology of Learning and Memory, 134, 178–191.Nadel, L. & Moscovitcht, M. (1997) Memory consolidation, retrograde amnesia and the hippocampal complex. Current Opinion in Neurobiology, 7, 217–227.Naqshbandi, M., Feeney, M.C., McKenzie, T.L.B., & Roberts, W.A. (2007) Testing for episodic-like memory in rats in the absence of time of day cues: Replication of Babb and Crystal. Behavioural Processes, 74, 217–225.Nonaka, M., Fitzpatrick, R., Lapira, J., Wheeler, D., Spooner, P.A., Corcoles-Parada, M., Muñoz-López, M., Tully, T., Peters, M., & Morris, R.G.M. (2017) Everyday memory: towards a translationally effective method of modelling the encoding, forgetting and enhancement of memory. European Journal of Neuroscience, 46, 1937–1953.Nyberg, L., McIntosh, A.R., Cabeza, R., Habib, R., Houle, S., & Tulving, E. (1996) General and specific brain regions involved in encoding and retrieval of events: what, where, and when. Proc. Natl. Acad. Sci. U.S.A., 93, 11280–11285.O’Keefe, J. (1976) Place Units in the Hippocampus of the Freely Moving Rat. Experimental Neurology, 51, 78–109.O’Keefe, J. & Krupic, J. (2021) Do hippocampal pyramidal cells respond to nonspatial stimuli? Physiological Reviews, 101, 1427–1456.Prescott, T.J., Camilleri, D., Martinez-Hernandez, U., Damianou, A., & Lawrence, N.D. (2019) Memory and mental time travel in humans and social robots. Phil. Trans. R. Soc. B, 374, 20180025.Prodan, A., Davies, H., Eneqvist, H., Mastroberardino, G., Wijayathunga, H., Wardlaw, K., & Morris, R.G.M. (2022) Memory recall: New behavioral protocols for examining distinct forms of context specific recall in animals. Neurobiology of Learning and Memory, 195, 107685.Roberts, W.A., Feeney, M.C., MacPherson, K., Petter, M., McMillan, N., & Musolino, E. (2008) Episodic-Like Memory in Rats: Is It Based on When or How Long Ago? Science, 320, 113–115.Scoville, W.B. & Milner, B. (1957) Loss of recent memory after bilateral hippocampal lesions. Journal of neurology, neurosurgery, and psychiatry, 20, 11–21.Squire, L.R. (1984) Mechanisms of Memory. Science, 232, 1612–1619.Steele, R.J. & Morris, R.G.M. (1999) Delay-dependent impairment of a matching-to-place task with chronic and intrahippocampal infusion of the NMDA-antagonist D-AP5. Hippocampus, 9, 118–136.Tse, D., Langston, R.F., Kakeyama, M., Bethus, I., Spooner, P.A., Wood, E.R., Witter, M.P., & Morris, R.G.M. (2007) Schemas and memory consolidation. Science, 316, 76–82.Tse, D., Privitera, L., Norton, A.C., Gobbo, F., Spooner, P., Takeuchi, T., Martin, S.J., & Morris, R.G.M. (2023) Cell-type-specific optogenetic stimulation of the locus coeruleus induces slow-onset potentiation and enhances everyday memory in rats. Proc. Natl. Acad. Sci. U.S.A., 120, e2307275120.Tulving, E. (1985) Memory and consciousness. Canadian Psychology, 26 .Tulving, E. (2005) Episodic Memory and Autonoesis: Uniquely Human? In The Missing Link in Cognition: Origins of Self-Reflective Consciousness . Oxford University Press.Wang, S.H., Redondo, R.L., & Morris, R.G.M. (2010) Relevance of synaptic tagging and capture to the persistence of long-term potentiation and everyday spatial memory. Proceedings of the National Academy of Sciences of the United States of America, 107, 19537–19542.Wilkinson, J.L., Herrman, L., Palmatier, M.I., & Bevins, R.A. (2006) Rats’ novel object interaction as a measure of environmental familiarity. Learning and Motivation, 37, 131–148.Zhou, W. & Crystal, J.D. (2009) Evidence for remembering when events occurred in a rodent model of episodic memory. Proceedings of the National Academy of Sciences of the United States of America, 106, 9525–9529. Information & Authors Information Version history Copyright This work is licensed under a Non Exclusive No Reuse License. Collection

Authors Metrics & Citations Metrics Article Usage 393views 207downloads Citations Download citation Kayleigh Kanakis, Richard G M Morris, Francesco Gobbo. Modelling of what-where-when episodic-like memories in rats. Authorea. 09 April 2025. DOI: https://doi.org/10.22541/au.174420658.84041878/v1 DOI: https://doi.org/10.22541/au.174420658.84041878/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu.