An investigation of social learning of tool-use in bottlenose dolphins (Tursiops truncatus) using a ball-up / ball-down task

15 16 Whether and how dolphins can engage in spontaneous, untrained social learning to solve 17 novel tool problems via action- or result- (or other) social learning remains debated. In the 18 present study we tested the spontaneous social learning abilities of six dolphins (all not 19 trained to copy) on two tool-using tasks (what we call the ball-up / ball-down task) - using a 20 dolphin demonstrator and a human demonstrator. Regardless of task type and demonstrator 21 type none of the tested dolphins reproduced the demonstrated tool solutions. We experienced 22 several issues regarding our test apparatus, and so these negative results may be due to 23 apparatus failures. However, these findings may also fully or partially indicate that untrained 24 dolphins are not generalized, spontaneous social learners across information types, especially 25 regarding the acquisition of tool solutions in puzzle tasks. More studies are required to 26 precisely determine the capacity for spontaneous tool solution copying in dolphins. 27 28 29

social learning, tool-use, problem solving, training, culture, tradition, bottlenose 30 dolphin 31 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 3

32 33 All cultures are (directly and/or indirectly) social enterprises – culture is dependent on the 34 presence of effects of some type(s) of social learning. Social learning refers to the acquisition 35 of information from or via another individual or its behavioral products (Heyes 1994). Social 36 learning can take place via multiple social learning types, which are not well understood on a 37 neurological level but are typically differentiated based on the nature of the information 38 acquired (see reviews, e.g. by Whiten and Ham 1992). A class of information that involves 39 information about the biomechanical actions involved in a behavior and/or the artifact or 40 artefact movements resulting from the behavior is know-how (where the know-how can be 41 sequentially and/or hierarchically organized, too; Tennie et al., 2017). Social learning of 42 know-how may trigger the release and/or development of latent know-how in observers, or it 43 may lead to copies of know-how that the observers would (absent the observation) have been 44 very unlikely to perform (Tennie et al. 2017). Subtypes of social learning of know-how are 45 imitation, where observers act similar to demonstrators following their observation (e.g. 46 Zentall, 2006). Another subtype is emulation, which can can refer to the reproduction of the 47 outcomes (or, more psychologically oriented: the goals) of a behavior (Tomasello 1990; see 48 Huang & Charman, 2005 for an overview of emulation’s subdivisions), but using one's owns 49 means. Emulation learning itself can take place via goal-emulation (adopting an inferred goal) 50 and/or affordance learning (learning about the physical properties of the environment and 51 relations among objects) and/or object movement reenactment (replicating what objects did, 52 i.e. how they moved; reviewed by Hopper, 2010). In all these cases, the distinction between 53 triggering and copying can be upheld as described (e.g. the subtype imitation can lead to a 54 trigger or to a copy of action patterns). 55 56 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 4 Humans can copy know-how even if this know-how is beyond potential individual reach and 57 such know-how copying has been deemed a necessary enabling condition in human cultural 58 evolution (Boyd & Richerson, 1996; Tomasello, 1999; Tennie et al. 2017). The existence of 59 know-how copying abilities in non-human animals, however, is a much more debated topic. 60 Claims of know-how copying abilities are particularly common for great apes and cetaceans 61 (especially odontocets, Kuczaj II & Yeater, 2006). Chimpanzees (Pan troglodytes) for 62 example have been suggested to have to acquire (thus, a claim for a need for copying) some 63 of their cultural repertoires (patterns of behavioral variation within a species that are not 64 directly correlated with genetic or environmental differences between populations; see e.g. for 65 some such claims inside Whiten et al. 1999, 2001); additional claims of know-how copying 66 abilities of chimpanzees are common in the primatological literature (e.g. reviewed in Whiten 67 et al. 2009). However, other studies have suggested, based on experimental data, that great 68 apes do not spontaneously develop marked know-how copying abilities outside of human 69 influence (Tomasello & Call, 1997; Tennie et al. 2012; Neadle et al. 2020; Tomasello et al 70 1993b). 71 Multiple studies attribute social learning, and also copying, of know-how – both via 72 imitation and emulation – to dolphins, particularly to bottlenose dolphins (Tursiops truncatus 73 and Tursiops sp.; reviewed by Kuczaj II and Yeater 2006; Herman 2002; Kuczaj et al. 2012). 74 Potential evidence comes from four sources: first, there are studies looking into copying vocal 75 know-how. Second, there are studies on synchrony in dolphin behaviour, which is often 76 linked to action know-how copying skills; third, several anecdotal reports have been linked to 77 type of know-how copying; fourth, experimental studies have reported results congruent with 78 action know-how copying in dolphins. Fifth, other experimental studies claim for results 79 copying in tool use tasks. Sixth, there are studies on social learning of tool use in the wild. We 80 will discuss these sources in turn. But first, we note that the claim that dolphins have some 81 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 5 social learning of some type(s) of information is not an extraordinary claim. Social learning in 82 animals is extremely widespread, and it would be surprising to find that dolphins are 83 incapable of any type. Indeed, as we will see, the below provides good evidence for some 84 types of social learning in bottlenose dolphins. 85 86 Vocal behavior 87 88 Evidence exist for social learning of vocal behavior, such as so-called signature 89 whistles (evidence in bottlenose dolphins: King et al. 2013). Particularly strong and controlled 90 evidence for the spontaneous social learning of vocal know-how in dolphins stems from a 91 study on wild Atlantic spotted dolphins (Stenella frontalis, Herzing et al. 2024). Several of 92 these wild dolphins spontaneously (albeit partially) copied several computer-generated sounds 93 (CGS) – including even the "start" and "stop" tones that preceded and followed the CGSs that 94 these dolphins were exposed to. These computer generated sounds (and even the start and stop 95 tones) were novel to there dolphins and would highly unlikely have been used or developed 96 by these dolphins lacking the demonstrations (as the authors state these sounds “were 97 designed to be outside the dolphin’s natural repertoire” and this was tested against baseline 98 data). The fact that these wild dolphins (though note that they had years of human contact) 99 nevertheless copied these sounds – absent human training and (possibly) absent any human 100 enculturation - renders the results of this study a relatively clear case of spontaneous vocal 101 know-how copying in dolphins. Thus, this is not only clear evidence of social learning in 102 dolphins, but also clear evidence for know-how copying. However, the ability to copy in one 103 domain (here the vocal domain) need not entail similar abilities in other domains. We will 104 thus turn to these other domains. 105 106 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 6 Synchronous behavior 107 108 Adult dolphins spend a considerable amount of time in synchrony with other dolphins 109 (Connor et al. 2000; Connor et al. 2006; Fellner et al. 2006; Perelberg & Schuster, 2008). 110 Why dolphins synchronize is not altogether clear but previous studies have suggested that the 111 coordinated breathing of dolphins represents an act of cooperation (Perelberg & Schuster, 112 2008), and that male bottlenose dolphins might surface synchronously as an alliance signal 113 (Connor et al. 2006). In some ways, even the fact that dolphins can form social groups is in 114 itself evidence for some social learning: the very location of dolphins in time and space is 115 clearly influenced by other dolphins – thus satisfying the minimal definition of social learning 116 given above. Moreover, however, dolphins frequently exhibit spontaneous synchronous group 117 behavior while traveling, foraging, playing, resting and displaying (Connor et al. 2000; 118 Connor et al. 2006; Fellner et al. 2006; Miles & Herzing, 2003). Furthermore, during the first 119 three months of a newborn's life, a mother spends 80% of her time in synchrony with the 120 newborn (Fellner et al. 2006; Mann & Smuts, 1999; Miles & Herzing, 2003). As social 121 behavior and synchronous behavior appear in wild dolphins, dolphins clearly perform 122 synchronous behavior spontaneously – in the sense of not requiring human influences to 123 develop this aspect of their behavior. It has even been suggested that imitation (or more 124 neutrally: social learning regarding actions) in dolphins might spontaneously develop first 125 from passive and later from active maintenance of such synchrony (Fellner et al. 2006). 126 Synchronous behavior in dolphins has also been tested in captivity, though after 127 training and human interference. Two captive bottlenose dolphins were human-trained to 128 perform a particular behavior in tandem (i.e., in synchrony) upon receiving a signal from their 129 trainer (Braslau-Schneck, 1994; as cited in Herman, 2002). Additionally, the same dolphins 130 were later signaled to perform an unspecified behavior in synchrony. In both cases, the 131 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 7 dolphins executed the behaviors in almost perfect synchrony (after both dolphins had first 132 spent several seconds swimming together underwater), with one dolphin always performing 133 slightly ahead of the other dolphin (Braslau-Schneck, 1994; as cited in Herman, 2002). 134 How exactly captive and wild dolphins achieve synchronization, remains debated. One 135 possibility could be that a slightly delayed dolphin copies (or gets triggered by) another 136 dolphin's actions or muscle innervations. However, it is not clear whether synchronous 137 behavior represents a case of this per se, as social learning about others’ actions might not be 138 a necessary requisite in order to synchronize – note also that not all muscle innervations are 139 easy to observe or infer (all this in a potentially moving medium of water). Synchronous 140 behaviors may instead occur as a consequence of the lagging dolphin closely following (in 141 parallel) the direction of movement (or a parallel position’s match) of another dolphin without 142 necessarily copying the leading dolphin’s actions or muscle innervations (that is, instead of 143 copying or triggering action know-how, dolphin synchrony may be due to social learning of 144 “know-where” another dolphin is across time (know-when) perhaps in relation to oneself; 145 compare Bandini et al. 2020, Tennie et al. 2020). Overall, the available data on synchrony 146 therefore does not unquestionably demonstrate a role of action / muscle social learning (be it 147 triggering or copying) in the performance of synchronous behavior and in particular it does 148 not demonstrate that observers copied bodily actions that they themselves could not have 149 developed on their own absent demonstrations. 150 151 Anecdotal observations 152 153 Anecdotal evidence for action social learning in dolphins includes instances of 154 dolphins (bottlenose dolphins and one false killer whale, Pseudorca crassidens) claimed to 155 reproduce actions performed by other dolphins (Brown et al. 1966; Caldwell et al. 1965; 156 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 8 Pace, 2000; Tayler & Saayman, 1973) as well as reproducing actions performed by 157 individuals of other species (humans, Cape fur seal, fish, penguins, skates and loggerhead 158 turtles, see Tayler & Saayman, 1973; Kuczaj et al. 2012; for a review see Herman, 1980). 159 Bossley et al. (2018) reported a case in which a wild dolphin was temporarily housed with 160 captive, trained dolphins. Once released, the dolphin started performing a behavior (tail 161 walking) that was present (human-trained) in the captive population but was never observed 162 before in the wild. Furthermore, years later, another wild dolphin started to perform tail 163 walking behavior. Although these observations provide evidence for the existence of social 164 learning in dolphins (especially the second wild tail walker, who presumably never had much 165 human contact), they do not provide conclusive evidence of the social learning of actions 166 (though they are consistent with this possibility). Namely, it is again unclear what kind of 167 information these two dolphins acquired from or via the demonstrators. Instead of copying the 168 actions, perhaps these dolphins learnt more generally that their bodies could be maintained 169 outside of the water (“know-where” information; or maybe the learning related to the goal of 170 moving the body out of the water) and they then subsequently reinnovated the behavioral 171 form itself (tail walking) on their own – e.g. by trial and error learning. 172 173 Whilst anecdotal evidence can be helpful insofar as it can guide the development of 174 experimental studies or more systematic observations, anecdotes can hardly be regarded as a 175 research strategy on their own (Maestripieri & Whitham, 2001) and systematic studies are 176 needed to test the hypotheses stemming from anecdotes. We therefore note that these 177 anecdotes are of interest and consistent with the idea of action social learning (and perhaps 178 even action copying), but they alone cannot prove this as too many alternative possibilities 179 remain. Yet, they once again show the presence of spontaneous social learning in wild 180 dolphins. 181 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 9 182 Experimental studies on actions 183 184 The fourth source of information about dolphin know-how social learning abilities 185

from experimental studies on action social learning. As alluded to already, 186 experimental studies on captive dolphins often have the limitation that the individuals may 187 not be ecologically representative of their wild counterparts because they may be human-188 enculturated (henceforth: enculturated) and/or human-trained. Data obtained from such 189 studies may not capture what dolphins may do spontaneously, absent human interference. 190 Enculturation refers to animals that have been reared in a human cultural environment with 191 wide exposure to human artefacts and/or social/communicative interactions (Furlong et al., 192 2008). Enculturation has been suggested to be able to principally induce cognitive abilities 193 that are not present and would not develop otherwise in wild, untrained conspecifics - such as 194 the ability to readily copy actions in great apes (Henrich & Tennie, 2017; Tomasello & Call, 195 2004; Tomasello et al. 1993b). 196 Two studies (Herman et al. 1989; as cited in Herman, 2002; Xitco, 1988) used the so 197 called "Do as I do" paradigm to investigate copying abilities in bottlenose dolphins. By 198 necessity, in the “Do as I do” paradigm, a subject is first human-trained (by shaping, for 199 example) to reproduce some demonstrated actions on command (the command often being the 200 words “Do this”, or an equivalent gestural signal). The subject is then tested on transfer 201 actions (which often also include the originally trained actions – but really should be novel, 202 and ideally, actions that would not otherwise occur). The study by Xitco (1988) involved two 203 bottlenose dolphins (Phoenix and Ake) who acted as demonstrator and as observer for each 204 other (Xitco, 1988). The transfer actions were either from the preexisting repertoire of both 205 dolphins (familiar actions - thus not fully improbable "transfer" actions) or they were 206 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 10 previously taught to only one dolphin - the demonstrator - and so were considered novel for 207 the other dolphin - the observer (i.e., these were assumed to be “novel” actions, but see Byrne 208 and Tanner, (2006) for a general critique of such an approach). Familiar actions were 209 reproduced. Phoenix successfully reproduced 7 of 12 of the familiar actions and Ake 210 successfully reproduced 6 of the 12 familiar actions. Phoenix also reproduced 2 of the 3 novel 211 actions (both on the second trial) and Ake reproduced 1 of 3 of the novel actions (on the third 212 trial). 213 A later study using the Do as I Do paradigm on the same two dolphins (Herman et al. 214 1989; as cited in Herman, 2002) also compared the repetition rate of actions depending on 215 whether the demonstrator was a dolphin or a human. The authors found that there was no 216 significant difference in action repetition by these two trained dolphins depending on whether 217 the demonstrator was a dolphin (62% correct) or a human (58% correct). Although these rates 218 of repetition might seem high, it is important to take into consideration that both dolphins 219 tested by Xitco (1988) and Herman et al (1989) had also experienced years of language-220 training where they were required to produce various actions on command and that they had 221 ample experience in performing actions in tandem on command. Given that these dolphins 222 received approximately eight to ten hours a day of human exposure (including training and 223 cognitive experiments), these individuals should be considered not merely human-trained but 224 also human-enculturated (Herman, 2002). 225 While the aforementioned studies show some potential under special circumstances 226 (human interference) for action copying in bottlenose dolphins, their ecological validity is 227 clearly compromised both by their degree of training and enculturation of the tested subjects. 228 It is possible, if not likely, that the detected action copying abilities might have been induced 229 in the two tested dolphins through extensive training and/or enculturation. 230 Bauer and Johnson (1994) set out to replicate Xitco (1988) study in order to test the 231 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 11 copying abilities of bottlenose dolphins who had not had years of prior specialized training. 232 The subjects in Bauer and Johnson's study (two bottlenose dolphins named Toby and Bob) 233 were tested using the same experimental procedure as Xitco (1988). Although Toby and Bob 234 learned most of the training actions and almost all of the transfer actions a few months prior to 235 the study in the Do as I Do paradigm, the two dolphins were described to have had worse 236 reproduction rates than those dolphins tested by Xitco (1988) when asked to replicate familiar 237 actions. Moreover, when asked to replicate novel actions, neither of the two dolphins 238 replicated any of the actions correctly (Bauer and Johnson, 1994). Based on these results, 239 Bauer and Johnson concluded (in the terminology of the current paper) that their two dolphins 240 lacked action copying skills. Consequently, it is possible that only dolphins that have 241 experienced enculturation and/or other, specialized training can copy – mere training (as in 242 Do As I Do training) may not suffice. Of course, potentially many other differences may 243 explain these discrepant results. Overall, it therefore remains an open question whether 244 dolphins can easily and spontaneously (learn to) copy know-how (here: action know-how) – 245 though the study by Bauer and Johnson (1994) suggests that they do not. What is remarkable 246 is that even following some human-training-to-copy, dolphins can still fail to copy novel 247 actions (Bauer and Johnson 1994). 248 249 Experimental studies on environmental results 250 251 Regarding the question of whether dolphins can copy environmental results (a variant 252 of emulation; see above), we are aware of only one such study that contained a claim of such 253 abilities (sensu Kuczaj & Walker, 2006). Captive bottlenose dolphins were reported to 254 spontaneously copy results achieved by human demonstrators in a problem-solving task. In 255 particular, in one of the experimental conditions called the "multiple-weight task", dolphins 256 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 12 had to pick up and later drop four weights into a container in order for a fish reward to be 257 released, which they allegedly learned by observing human divers' results-demonstrations 258 (sensu Kuczaj & Walker, 2006). Given that the target of the task was to place the weights 259 (know-what; Tennie et al. 2020) in a precise location (know-where; Tennie et al. 2020) rather 260 than to investigate if the dolphins could perform the actions involved in solving the task 261 (which they were already trained to do, Kuczaj et al. 1998), reproduction of the solution 262 demonstrated by the divers could evidence emulative abilities in dolphins in the sense of 263 know-how, or else the social learning of a combination of know-what and know-where (see 264 Tennie et al. 2020). In addition, as the task was not novel to the dolphins, it is not possible to 265 rule out carryover effects from previous experiments – meaning the novel, unlikely know-how 266 transmission can be ruled out (not a test for results copying). 267 In another experimental condition of the same study (sensu Kuczaj & Walker, 2006) 268 called the "2-Step Time-Limited task", the same dolphins that participated in the previous 269 condition had to now use two tools (a weight and a stick) in quick succession (15 s) in order 270 to get to a fish reward out of a puzzle box (sensu Kuczaj & Walker, 2006). Again, the authors 271 mention (but do not elaborate) that dolphins learned to use these tools by observing a human 272 demonstrator. Later during the experiment, the stick-tool was placed further away from the 273 tool site and the dolphins had to bring it closer to the apparatus before they used the weight in 274 order to also be able to use the stick within the limited time frame. None of the dolphins 275 spontaneously brought the stick close to the puzzle before using the weight box. However, the 276 authors report that after one of the dolphins saw a human model bringing the stick to the 277 puzzle box before using the weight, one dolphin "quickly began to do so himself" (sensu 278 Kuczaj & Walker, 2006). What, if and how the dolphins actually learned to solve the task is 279 again unclear – especially given that the amount of training they had received is not described 280 in detail. Know-how copying may be a possibility, in the sense of results copying. But other 281 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 13 types of social learning may have acted instead or in addition (such as social learning of 282 know-what and know-where and know-when). Overall, then, none of the studies to date 283 conclusively show that dolphins spontaneously copy either novel actions or novel results 284 (know-how copying) – but tool use tasks (such as the weight or weight and stick task) seem to 285 hold some promise. In general, social learning can take place across several behavioral 286 domains but in the present study, and from here on, we shall focus on the technological (here: 287 tool) domain. 288 289 Tool use in wild dolphins 290 291 Some individuals in populations of bottlenose dolphins have been described to use 292 tools in foraging contexts. One tool use behavior described in bottlenose dolphins, sponging, 293 was reported in Shark Bay (Australia) (Krützen et al. 2005; Mann et al. 2008; but see also 294 Parra, 2007 for observations on one sponge carrying Indo-Pacific humpback dolphin). 295 Sponging consists in the use of conical sponges likely as "gloves" for the rostrum while 296 foraging for buried prey in the sand. Sponging behavior is highly biased towards females and 297 recent studies have shown that genetic and environmental factors do not account on their own 298 for the pattern of distribution of this behavior within matrilines, suggesting that (vertical) 299 social learning of some type(s) is present in sponging behavior (Wild et al. 2019; Krützen et 300 al. 2005). This leaves open the precise social learning type(s) that may be involved. A second 301 tool use behavior described in wild bottlenose dolphins is shelling (Allen et al. 2011). During 302 shelling, dolphins guide prey into empty gastropod shells or directly feed on prey hiding in 303 such shells by carrying the shells to the surface, emptying the water and shaking the shells 304 (Allen et al. 2011). Recent analysis again incorporating both genetic and environmental data 305 have shown that contrary to sponging, shelling, too provides evidence for some social 306 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 14 learning. In shelling, the social learning was found not to be vertical but rather to horizontal or 307 oblique, as in social learning between peers (Wild et al. 2020). Again, this leaves open the 308 social learning type(s) that may be involved. 309 There are more types of tool use in dolphins than the two described (e.g. among them 310 so-called mud-ring feeding (Torres & Read 2009, Engleby & Powell 2019) and also social 311 tool use in cooperation settings – even across species (Simões-Lopes et al. 1998). 312 Unfortunately, none of the observations of tool use in wild dolphins allow for easy assessing 313 which social learning types accompany, underlie or perhaps have to underlie (the latter a 314 claim for copying) the acquisition of these tool-based foraging techniques. Given that the 315 action social learning capacity of dolphins has been mainly addressed using demonstrations of 316 behaviors already within the dolphins’ repertoire of bodily movements, and that the tool use 317 tests performed in captivity are inconclusive (see above), we may say that "Social learning of 318 relatively novel techniques of food-handling […] has yet to be demonstrated." (Whiten & van 319 Schaik, 2007, pg. 615) using " well-controlled experiments" (Mann et al. 2007; Sargeant et al. 320 2005) – in the sense of the techniques themselves being socially transmitted. 321 A firm answer to what type(s) of social learning may or may not be spontaneously 322 present in dolphin tool behavior – and especially which type(s) may be necessarily required – 323 is far off at the present time. However, to provide better and firmer answers, more studies are 324 required, that systematically vary information types in demonstrations to determine which 325 types of information are and which are not (or to a lesser degree) socially learned by dolphins. 326 As one additional important question related to the presence of absence of know-how copying 327 types (as defined above), such studies should also ideally test the baseline competencies in 328 motivated subjects. In all cases, to attain ecologically and phylogenetically meaningful 329 answers, tests should be conducted that test for spontaneous or spontaneously developing 330 abilities – that is, untrained and ideally unenculturated subjects should be tested. 331 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 15 The present study was designed to test for the spontaneous social learning abilities of 332 two types of tool-use behaviors by bottlenose dolphins (Tursiops truncatus). We conducted 333 these two studies with bottlenose dolphins who had mild levels of enculturation (at least when 334 compared to some of the earlier tested dolphins) and who had also not been trained on any of 335 the actions or results needed to solve the target tasks. We adapted parts of the experimental 336 design of the "multiple-weight task" used by Gory and Kuczaj (sensu Kuczaj & Walker, 337 2006) to simplify the task given the lack of preliminary training in the tested dolphins. The 338

we label the ball-up / ball-down task. Subjects were required to release only one 339 "weight" (here: a ball; this being the know-what) - instead of four - in order to obtain a fish 340 reward. In the first study (the ball-up variant of the task), the dolphins had at their disposal 341 air-filled balls (know-what 1) that could be used to retrieve the fish by releasing them at the 342 lower end of the tube (i.e., the ball rising in the tube would displace the fish out the other end; 343 know-where 1). In the second study (the ball-down variant), the dolphins were given 344 negatively buoyant balls (know-what 2) that could be placed in the top of the tube (know-345 where 2) thereby having them sink through the tube and displacing the fish at the bottom. In 346 the first task the demonstrator was a dolphin (who demonstrated target results as well as target 347 actions), and in the second task the demonstrator was a human (who modeled target results 348 but not target actions, since she used her hands). We report on these two studies now, 349 although we note that we experienced some methodological issues with our setting, which 350 prevent conclusive claims being drawn. 351 352 Study 1 – Ball up (dolphin demonstrations) 353 354

355 356 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 16 Subjects 357 Subjects were six bottlenose dolphins (Tursiops truncatus) housed in the dolphinarium at the 358 Tiergarten Nürnberg (Germany) (details Table 1). The dolphins were kept together with six 359 California sea lions (Zalophus californianus) in three indoor tanks connected to each other 360 through short passages. The experimental sessions took place in a circular tank 12m in 361 diameter and 4m in depth (from now on called the "testing tank"). Its adjoining tank (25m x 362 11m, 4.7m deep) was used for public performances and training. The third tank (9m x 5m, 363 2.5m deep) was on the other side of the "performance" tank. The dolphins were fed five to six 364 times a day and were never food deprived during the study. At the time, the dolphins 365 participated in four to five trainings or public performances a day, depending on the season. 366 One dolphin, Noah, had previously participated in two studies on numerical competence 367 (Kilian et al. 2003; Kilian et al. 2005). 368 369 Table 1. Dolphins that participated in Study 1 and 2. 370 Dolphin Age (years) Born where In Nürnberg since Sunny 7 Soltau September 2005 Naomi 8 Nürnberg birth Noah 13 Nürnberg birth Jenny 21 wild 1991 Eva 37 wild 1979 Moby 46 wild 1971 Note. Ages for wild born dolphins are approximate. 371 372 Apparatus 373 The testing apparatus was a clear hollow polycarbonate transparent tube (90cm long and 374 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 17 23cm in diameter) open on both ends. This apparatus was baited with a fish inside and 375 attached vertically to a metal frame at the edge of the testing tank. The fish-bait was attached 376 in the middle of the tube with a white plastic lacing cord. The shallowest opening of the tube 377 was fixed at a depth of 30cm from the water surface (see fig. 1). 378 The "tools" provided to the dolphins were small air-filled basketballs (know-what 1; 379 15cm in diameter). The dolphins knew these items, as they were often used during 380 trainings/performances by throwing them out of the pool to the trainer or balancing them on 381 their rostrum. The dolphins also played with such balls extensively outside of training 382 sessions by carrying them in their mouth, throwing them above the water or out of the tank or 383 swimming under water with them. 384 In the demonstration phase for Group 2 (see Procedure section), a target was fixed to 385 the tube. A target is a regular training tool used to shape an animal's behavior. In this case the 386 target was a stick (130cm) with a tennis ball attached to the end vertically fixed to the outside 387 of the tube and taped at both ends so that the tennis ball rested just below the bottom lip of the 388 tube. 389 Before the study started, we verified whether the dolphins could perceive the 390 (transparent) tube by throwing a similar tube (smaller in diameter and length but of the same 391 material) into the testing tank. Once the tube reached the bottom of the tank, the trainer 392 instructed the dolphins to "bring the object back". Given that the tube was immediately 393 retrieved by the dolphins, we assumed that the tube was visible and could be perceived by the 394 dolphins. The dolphins could also clearly perceive the fish in the tube, as was apparent from 395 the fact that when they were let for the first time into the testing tank with the baited apparatus 396 present (baseline phase), they swam directly towards the fish and tried to extract it. 397 398 399 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 18 Fig. 1a 400 401 402 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 19 Fig. 1b 403 404 Figures. 1a and 1b. Apparatus. To get the fish-reward in Study 1 (Ball up) the subjects had 405 to position the air-filled ball (know-what 1) under the tube's lower opening (know-where 1) 406 and let it go. The ball then rose to the water surface and untangled the fish from the lacing 407 cord on the way up. In Study 2 (Ball down) they had to bring the heavy ball (know-what 2) 408 over the tube's upper opening (know-where 2) and let it go. The heavy ball then sank 409 through the tube to the tank's floor and again untangled the fish and brought it with it when 410 it came down. Note Figure 1a is not to scale. 411 412 Testing Procedure 413 One above-water camera was positioned at the edge of the pool some meters away from the 414 apparatus in order to film both the apparatus and the area around it (approximately 2 meters in 415 diameter). 416 Study 1 consisted of four phases: 1.) baseline, 2.) training of a dolphin demonstrator, 417 3.) target control, and 4.) trials with demonstration. The baseline phase was conducted with all 418 the dolphins together. In the demonstrator training phase, the dolphin acting as demonstrator 419 was separated from her group. In the target control and the trials with demonstrations the 420 dolphins were divided into two groups. Group 1 comprised of Moby, Eva and Sunny and 421 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 20 Group 2 of Noah and Naomi. We tested the dolphins in groups instead of individually, 422 because the dolphins were not used to being separated from the group (except for medical 423 reasons). 424 1.) Baseline. The baseline phase consisted of five 15 minutes sessions and was 425 designed to familiarize the subjects with the experimental apparatus and to document the 426 actions and choices displayed spontaneously by the subjects (including spontaneous 427 occurrences of the use of the balls as tools). At the beginning of each baseline session, the 428 baited apparatus and six balls were placed in the water before the dolphins were let into the 429 testing tank. This baseline acted as a test for whether the dolphins required copying to solve 430 the task. 431 2.) Training of the dolphin demonstrator. Jenny was used as the dolphin demonstrator, 432 and so she was trained. For the time of training she was separated from the rest of the group. 433 We used targeting to train Jenny to use the ball as a tool to get the fish out of the tube by 434 letting the ball float upwards from the lower tube opening. Targeting is a shaping technique 435 where an animal is taught to touch some part of its body with an object, called the target 436 (Ramirez, 1999). The target was used to guide Jenny toward the tube's lower opening. At the 437 beginning of the training phase, Jenny was given a ball and a new hand signal, and then she 438 was guided by the target to the lower hole of the tube. The training phase consisted of 10 439 sessions, each with 10 to 16 trials. 440 3.) Target control. Given that Jenny needed the guidance of a target in order to be a 441 successful demonstrator, we added one control session to control for possible effects of the 442 target on the dolphins’ behavior toward the tube (as compared to the baseline). The procedure 443 was exactly the same as in the baseline sessions and each session (1 per group) lasted 10 444 minutes. For Group 1 the experimenter held the target in the exact same position as she held it 445 for Jenny during the demonstrator training phase but for Group 2 the target was fixed to the 446 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 21 tube. 447 4.) Trials with demonstrations. Jenny served as the demonstrator for both groups 448 during four sessions in each group. Each demonstration started by giving Jenny a ball and the 449 hand signal to start the target behavior (bring ball to lower opening of the tube and release the 450 ball). During the demonstrations the dolphins could freely swim around the testing tank. For 451 Group 1 a session consisted of four to six consecutive successful demonstrations – then 452 followed by a trial. For Group 2 a session consisted of three consecutive successful 453 demonstrations - then followed by a trial. For Group 2, this was succeeded by two additional 454 demonstrations, where each was followed by a trial. At the end of the session, both groups of 455 dolphins got one more demonstration without a consecutive trial. After each demonstration 456 the ball and the tube were taken out of the tank and the tube was re-baited out of sight of the 457 dolphins. While re-baiting took place the dolphins in both groups were distracted by being 458 asked to perform some training actions. During the experimental trials in phase 4, Jenny was 459 taken to the side of the tank and received medical training (i.e., lying still on her back). The 460 trials for Group 1 were approximately three minutes long and the trials for Group 2 were 461 approximately four minutes long. Trials normally followed demonstrations with a delay of 462 two to five minutes for Group 1 and one to two minutes for Group 2. 463 If Jenny was successful in retrieving the fish during a demonstration of the target 464 behavior, an experimenter blew a whistle (a secondary reinforcement used in dolphin 465 training). If Jenny performed an unsuccessful demonstration, she was given the signal to 466 repeat the task. Jenny’s performed successful demonstrations 48% of the times (23 times in 467 Group 1 and 24 times in Group 2). Unsuccessful demonstrations were often the result of 468 Jenny not swimming deep enough to insert the ball because she swam directly to the fish in 469 the middle of the tube. However, whenever Jenny used the apparatus correctly - i.e., whenever 470 she took the ball and then swam to the bottom of the tube - her success rate was 100%. 471 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 22 While Jenny ate the fish every time she completed a successful demonstration for 472 Group 1, in Group 2 another dolphin (Noah) stole the fish every time she performed a 473 successful demonstration. To prevent Jenny from getting frustrated, she was manually 474 rewarded by the experimenter after every successful demonstration. Given that in Group 2 475 Noah always took the fish after Jenny solved the task, it is difficult to say how much and what 476 exactly Noah and Naomi learned from the demonstrations. To motivate the members of 477 Group 2 to solve the task themselves without relying on Jenny, an extra 15 minute trial was 478 conducted after the fourth demonstration session in phase 4. 479 480 Coding 481 During phase 4 (trials with demonstrations), two experimenters next to the camera loudly 482 described which dolphins approached the apparatus and what they were doing. One of the 483 experimenters later coded the following variables from the video recordings: in phases 1 and 484 3 (baseline and trial control) the experimenter coded the number of times the dolphins 485 approached the apparatus with the ball. An approach was coded when the dolphin (with a 486 ball) came into a radius of 1 meter from the apparatus while holding the ball in the rostrum (a 487 rough measure of know-what and know-where). Form each approach the direction of the 488 approach was coded: from below – the subject approaches the bottom of the tube; from the 489 middle - the subject approached the middle of the tube where the fish was attached (a natural 490 distractor); and from above – the subject approached the top of the tube swimming in the 491 surface. From each approach we also coded if a ball was inserted in the tube and from where 492 was the ball inserted (know-where 1 or 2). 493 Unfortunately, it was not possible to confidently code from the video recordings if the 494 observer dolphins in each group had seen the demonstrations or not - given that dolphin's 495 vision field is maximized laterally. 496 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 23 497

and discussion Study 1 498 499 None of the dolphins solved the task neither in the baseline nor after seeing demonstrations 500 (in phase 4). There were no statistical differences in the individual number of approaches to 501 the apparatus between the baseline and the trial demonstration phase (paired Wilcoxon test: z 502 = 0.365, P = 0.875, N = 5). In the baseline Noah and Naomi were the only dolphins that 503 approached the tube with the ball – thereby showing that ball approach per se does not 504 necessitate demonstrations. In the demonstration phase four out of five dolphins approached 505 the tube with the ball at least once – perhaps a mild social learning effect. However, the 506 dolphins always approached the tube from above (Table 2). In the target control phase, the 507 dolphins occasionally touched the target with their rostrum but none of them approached the 508 tube or the target with a ball. In the extra 15 minutes trial conducted with Group 2, Naomi 509 never approached the tube with the ball whereas Noah did so twice (from above). Thus, the 510 members of Group 2 approached the apparatus even less often when Jenny was not present in 511 phase 4 than during the trials where Jenny was acting as demonstrator. Therefore, we believe 512 that the failure of the dolphins in Group 2 to solve the task was not due to their lack of 513 motivation nor due to their reliance on Jenny to provide them with a fish. 514 Noah and Naomi appeared to try to solve the task in a different way than the 515 demonstrated solution: by hitting the tube with their bodies. Their strategy did not involve 516 using any tool and this alternative solution was used already during the baseline phase (thus, it 517 was spontaneous). Noah and Naomi kept hitting the tube throughout the baseline phase and 518 the trials with demonstrations. In the baseline phase they never got the fish out by hitting the 519 tube but in the trials with demonstrations they were successful, on three occasions: twice 520 while the tube was being put into the water and once during a trial. Unfortunately, it was not 521 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 24 possible to count the number of hits of the tube by each dolphin given that no underwater 522 camera was available. 523 524 Table 2. Frequency of approaches (per 60 minutes) with the ball to the apparatus during the 525 baseline phases and demonstration phases (trials) in Study 1 (Ball up) and Study 2 (Ball 526 down) studies. In the parentheses are the total numbers of approaches for each dolphin. 527 Subject Study 1 - Ball up Study 2 - Ball down Baseline Demo Baseline Demo Moby 0 13.3 (3) 7.5 (4) 5.4 (6) Eva 0 4.4 (1) 0 0 Jenny 0 / 1.9 (1) 17.2 (35) Sunny 0 0 0 0.7 (1) Noah 2.4 (3) 1.6 (1) 0 3.5 (6) Naomi 15.2 (19) 3.2 (2) 1.9 (1) 0 528 529 530 Study 2 – Ball down (human demonstrations) 531 532

533 534 Subjects 535 Subjects were the same as in Study 1. 536 537 Apparatus 538 The apparatus was the same as in Study 1 with the exception that circles (3cm in diameter) 539 were drawn randomly over the tube, and both ends of the tube were marked with silver tape to 540 increase their visibility. During this second study, no target was attached to the tube. The 541 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 25 same small basketballs (15cm in diameter) were provided as potential tools but now they were 542 filled with saturated salt water so that they sank to the bottom of the tank (know-what 2). The 543 intended (target) solution now involved placing these heavy balls into top of the tube (know-544 where 2), so that their sinking would release the fish reward. The dolphins were allowed to 545 play and get acquainted with these balls for 10 days prior to Study 2. The balls were left in the 546 tank on average 4 to 5 hours a day and sometimes they were left there overnight. 547 548 Experimental procedure 549 Study 2 consisted of two phases: 1.) baseline and 2.) trials with human demonstrations. 550 1.) Baseline. The baseline phase was conducted with the whole group and consisted of 551 two sessions of 15 minutes each. The goal of the baseline phase was to document the actions 552 that the dolphins displayed spontaneously toward the baited apparatus. As before, the 553 dolphins could swim freely in and out of the adjacent performance tank. At the beginning of 554 each session six sink balls were placed in the testing tank together with the baited apparatus. 555 2.) Trials with human demonstrations. This test condition included twelve sessions in 556 which human demonstrations were provided to the dolphins. In each session dolphins were 557 exposed to three consecutive demonstrations before participating in a trial where they were 558 allowed to interact with the testing materials. Then they got one more demonstration followed 559 by another trial. During the demonstrations, a human sat at the edge of the tank and released 560 the ball under water into the tube's upper opening, so it sank down the tube. In 30% of cases, 561 the sinking ball did not release the fish and the demonstration was repeated. After each 562 demonstration the ball was left in the tank. Trials were initiated one to two minutes after 563 demonstrations ended. 564 In the first six sessions all six dolphins were tested together while having free access 565 to all tanks. In these six sessions, Naomi often drew Jenny away from the tube. Since Jenny 566 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 26 had been the subject with the most approaches to the tube with a ball (and therefore the most 567 promising individual), we decided to take Naomi out of the experiment. However, it was not 568 possible to separate Naomi from the rest of the group and for the next three sessions (7,8,9) 569 Jenny and Sunny were tested alone. As Sunny seemed to present signs of distress in session 570 10 she was exchanged for Noah. In the last three sessions (10, 11 and 12) Jenny and Noah 571 were separated from the other dolphins in order to watch the demonstrations. Due to these 572 complications, the dolphins received different numbers of demonstrations: Moby, Eva and 573 Naomi were exposed to 20 demonstrations, Sunny to 29, Noah to 30 and Jenny to 39. In the 574 last six sessions (7,8,9,10,11 and 12) the dolphins were required to perform some training 575 actions while the tube was being re-baited. 576 577 Coding 578 In Study 2 the same coding scheme as in Study 1 was used. In Study 2 the ball stayed in the 579 tank after the first demonstration and so the dolphins already had the opportunity to interact 580 with the ball during the demonstrations as well as during the trials. Thus, coding started 581 already during the demonstrations. 582 583

and discussion Study 2 584 585 Despite having ample opportunities, no dolphin solved the task with the demonstrated 586 solution either in the baseline or the trials with demonstrations. In the baseline phase three 587 dolphins approached the tube with the ball (all from above) and in the demonstration phase 588 four dolphins approached the tube with the ball at least one time (Table 2). We thus did not 589 find significant differences in the number of approaches to the tube with the ball between the 590 baseline and the trials with demonstrations (paired Wilcoxon test: z = 0.405, P = 0.812, N = 591 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 27 6). During the trials with demonstrations, Moby, Sunny and Noah approached the tube from 592 above, whereas Jenny approached the tube mostly from below (13 times) and from the center 593 (21 times). On one occasion Jenny slid the ball upwards from the bottom to the top of the 594 tube's outside and then released the ball. Twice Jenny approached the tube from below and 595 then tossed the ball up. Jenny also frequently hit the tube with her rostrum - both while 596 holding the ball and sometimes without the ball. She banged the tube for the first time in the 597 second session, but it was not until session 5 that she started hitting it regularly (more than 15 598 times per session). She successfully retrieved the fish one time in this manner (during a trial 599 in session 7). Noah and Naomi successfully acquired the fish eight times with this alternative 600 banging technique in the first six sessions (hitting the tube ca. 10 to 15 times per session). 601 When, later, Noah was alone with Jenny, he was never successful again. Noah and Naomi 602 were mostly successful with the banging method (5 out of 8 times) before a trial started, that 603 is, while the tube was being put in the water, but they almost never hit the tube during 604 demonstrations. The occasional obtention of the fish using this alternative technique may well 605 have affected the degree of attention paid to the demonstrations and/or reduced the need to 606 learn a new solution for the task. 607 608 Comparing the results from both Ball up and Ball down studies 609 610 Dolphins' approaches to the apparatus with the ball during both studies are presented in Table 611 2. The number of times that individual dolphins approached the apparatus with the ball did 612 not differ between the two studies, neither in the baseline phase (paired Wilcoxon test: z = 613 0.365, P = 0.875, N = 6) nor in the trials with demonstration (paired Wilcoxon test: z = 614 1.1214, P = 0.312, N = 5). 615 616 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 28 General discussion 617 618 In this study we aimed to test the abilities of six mildly enculturated dolphins to solve a tool 619 using task either spontaneously during baselines or by socially learning the solution from or 620 via a demonstrator (either a conspecific or a human). None of the six dolphins solved the task 621 using demonstrated tools in the demonstrated way (using the correct know-what as a tool in 622 the correct know-where). The failure of the dolphins in this task is unlikely to be the 623 consequence of the tool use actions or results being too complicated for dolphins. In Study 1 624 all the subjects played with the air-filled balls and displayed actions with the balls very 625 similar to those needed to retrieve the fish (e.g. submerging with the ball and letting it go 626 again). However, the dolphins did not transfer these actions to the experimental paradigm, 627 they failed to connect the correct know-what to the correct know-where. Given that in Study 1 628 the air-filled ball needed to be introduced in the tube from below in order to obtain the fish, 629 the fact that most dolphins (except Jenny) always approached the tube from above suggests 630 that they did not understand the task itself. Overall, across two studies we found no evidence 631 of spontaneous tool use social learning abilities in untrained, mildly enculturated bottlenose 632 dolphins. 633 The disparity between our negative results and at least some of the previous literature 634 on dolphin social learning might however solely be the consequence of methodological 635 difficulties (more on this below). They may also reflect the different levels of 636 training/enculturation of the subjects included in the various studies. The test dolphins in our 637 study were not trained to copy demonstrated actions nor received training to solve the target 638 task in any of the two solution types, consistent with the possibility that untrained, mildly 639 enculturated dolphins do not (or rarely) express spontaneous copying abilities in the tool 640 domain. Experience and test-sophistication - especially in some way of know-how copying 641 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 29 (which the dolphins we tested did not have) – could thus be important factors in eliciting 642 know-how copying abilities in dolphins’ tool use tasks. However, this possibility is not the 643 only one. 644 Another difference between our study and the previous studies reporting copying in 645 dolphins that might have influenced our results is the time delay between the demonstrated 646 solution and the observers' reaction. In the present study, the dolphins were allowed to 647 participate in the experiment one to three minutes after the demonstration (and sometimes 648 even later). However, in previous studies (Kuczaj & Walker, 2006) the tested dolphins were 649 allowed to interact with the testing materials immediately, i.e. while the demonstrations took 650 place (Xitco’s, 1988). Supporting time delay as an explanation, the performance of the 651 dolphins tested by Xitco (1988) got worse when a delay was imposed between the time the 652 demonstrations took place and the time the dolphins could participate in the task. Future 653 studies could systematically investigate the potentially negative effect of time delays in 654 dolphin social learning abilities. 655 However, the tested dolphins in our study might simply have been unsuccessful 656 because they sometimes saw unsuccessful demonstrations (a problem especially for Study 1), 657 or because they did not understand the task. Horner and Whiten (2007) suggested that if apes 658 do not have a representation of the requirements of the task (i.e., if they do not understand the 659 causality of the required actions), they then fail to perform a (demonstrated) solution to a 660 problem. The same could be argued for the dolphins in our study. Our dolphins may not have 661 understood the connection between the ball going through the tube and the release of the fish. 662 Note that even the dolphin Jenny may not have fully understood the problem in the right way, 663 since (even though she had been trained to solve the task in Study 1) she was unable to 664 transfer this knowledge to Study 2. While this is a real possibility, note that a causal 665 understanding of the task would have also removed the need for the dolphins to copy. Had 666 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 30 they understood the causal demands and the causal demands alone, then they could have even 667 independently arrived at the solutions, irrespective of demonstrations. 668 Another possible explanation why the dolphins did not solve the task could be that the 669 objects the dolphins were required to use as tools in Study 1 were extensively used by the 670 dolphins in different contexts (such as playing). Balls were not novel to the tested dolphins. 671 As a result, it is possible that the dolphins’ performances suffered from so-called functional 672 fixedness. Functional fixedness “occurs when the priming of a conventional / regular use for 673 an artifact (i.e., here: air-filled balls) makes it difficult to envision task-relevant atypical uses 674 of the artifact“ (Barrett et al. 2008). To the dolphins, the air-filled balls' fixed function – 675 though perhaps achieved solely by personal experience instead of conventions 1 - was 676 primarily to play. This may have cognitively prevented the dolphins from seeing the balls' 677 potential new function as a tool (see also Hanus et al. 2011). However, note that we partially 678 accounted for this possibility, because, in Study 2 the dolphins had to solve the task using an 679 ball object with novel, unfamiliar properties: salt-water filled balls (heavy balls; know-what 680 2). Since these changes had no effect in the performance of the dolphins this goes somewhat 681 against the hypothesis that the original hindrance of the dolphins’ performance in Study 1 had 682 been functional fixedness. Of course, even despite different fillings, the objects used in Study 683 1 and 2 might still have been perceptually too similar to overcome potential functional 684 fixedness. 685 Last but not least, and as alluded to above, there were methodological shortcomings of 686 our study and which could have been fully or partly responsible for our failure to find tool use 687 social learning types in untrained, mildly enculturated dolphins. Our apparatus could be 688 1 Note that functional fixedness can be potentially established via different pathways . Fixedness may derive from purely personal experience or may be purely culturally/observationally based (note the term ”conventional“ in the quote above). In the human case, the two types could potentially mix so that individual experience may accompany cultural practices and vice versa. Here we target functional fixedness based on personal experience alone. .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 31 solved by the dolphins in unintended ways, and indeed this proved distractive. We hope future 689 studies can address these shortcomings. Such studies should ensure the sturdiness of any used 690 tool-apparatuses to prevent dolphins from being able to obtain the reward by simply hitting 691 the apparatus. Lacking a need to use a tool in our task designed to elicit differential tool use 692 (or at least differential know-what/know-where social learning) therefore renders our results 693 and conclusions tentative. Motivation is a pre-requisite for meaningful tests, and here, this 694 motivation was reduced by the possibility to access food rewards without tools – which was 695 additionally distracting from learning the intended way(s) to solve the task. 696 More changes in future studies would benefit the robustness of outcomes. First, if 697 possible, individual testing should be implemented to control the learning opportunities of 698 each dolphin separately and to prevent the dolphins from influencing each other's responses. 699 Second, future studies should employ underwater cameras to capture the responses of the 700 dolphins in detail – lacking these details curbed the conclusiveness of our findings. Third, to 701 further investigate the effect that prior training might have on the social learning abilities of 702 dolphins, the performance of trained and untrained dolphins should be compared (ditto for 703 levels of enculturation). Fourth, to evaluate the effect that the type of demonstrator has on the 704 performance of the dolphins, demonstrations from different species (conspecifics and non-705 conspecifics) should be provided within a single task. Fifth, age effects could be evaluated by 706 comparing the copying abilities of both calves and adults (compare Kuczaj et al. 2012). 707 Once again, we do not hold the view that dolphins are incapable of 708 spontaneous social learning. Most animals seem spontaneously capable of some social 709 learning – it would have been instead highly surprising and unusual if dolphins had been an 710 exception (above we showed that they are not: dolphins are clearly capable of some social 711 learning). Yet, the fact that dolphins can learn socially in some ways does not mean that 712 dolphins can spontaneously socially learn in all ways or that their social learning – for all 713 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 32 transmitted types of information – always goes beyond what the individual dolphin could 714 have discovered in their own lifetime if sufficiently motivated and in the right circumstances. 715 This includes the special case of know-how copying in the tool domain. We hope that future 716 studies will continue to investigate (using experimental controls) the tool use (and other) 717 social learning abilities of unenculturated and untrained dolphins and their relative depth - 718 also to better understand how tool use behaviors such as shelling or sponging may likely be 719 transmitted in wild dolphin populations. 720 721 Acknowledgments 722 723 We foremost thank Annette Kilian, who has sadly passed away after co-writing the first drafts 724 of this manuscript. Due to subsequent revisions, she cannot be held to all the views expressed. 725 We would like to express many thanks to Josep Call for his support, without whom this study 726 would not have taken place. We thank Nathan Pyne-Carter and Alba Motes Rodrigo for 727 improving and editing earlier manuscripts. In particular we are indebted to Tiergarten 728 Nürnberg for letting us conduct this research and all the dolphin keepers for their help, 729 especially to the head dolphin keeper Armin Fritz. We further thank Dr. Lorenzo von Fersen 730 and Raik Pieszek. 731 732 Note on this manuscript 733 734 Note that this is likely the final manuscript, as we are currently (at time of uploading) no 735 longer intending journal submission. This is because one of the authors has sadly passed away 736 (AK) and another author (AH) is currently no longer active in academia. 737 738 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 33

739 Allen SJ, Bejder L, Krützen M. 2011. Why do Indo-Pacific bottlenose dolphins (Tursiops sp.) 740 carry conch shells (Turbinella sp.) in Shark Bay, Western Australia? Marine Mammal 741 Science 27: 449-454. 742 Bandini E, Motes-rodrigo A, Steele MP, Rutz C, Tennie C. 2020. Examining the mechanisms 743 underlying the acquisition of animal tool behaviour. Biology Letters 16: 20200122. 744 Barrett CH, Laurence S, Margolis E. 2008. Artifacts and Original Intent: A Cross-Cultural 745 Perspective on the Design Stance. Journal of Cognition and Culture 8: 1-22. 746 Bauer GB, Johnson CM. 1994. Trained Motor Imitation by Bottlenose Dolphins (Tursiops 747 truncatus). Perceptual & Motor Skills 79: 1307-1315. 748 Biro D, Inoue-Nakamura N, Tonooka R, Yamakoshi G, Sousa C, Matsuzawa T. 2003. 749 Cultural Innovation and Transmission of Tool Use in Wild Chimpanzees: Evidence 750 from Field Experiments. Animal Cognition 6: 213-223. 751 Bossley M, Steiner A, Brakes P, Shrimpton J, Foster C, Rendell L. 2018 Tail walking in a 752 bottlenose dolphin community: the rise and fall of an arbitrary cultural ‘fad’. Biology 753 Letters 14: 20180314. 754 Boyd R, Richerson PJ. 1996. Why culture is common, but cultural evolution is rare. 755 Proceedings-British Academy 88: 77-94. OXFORD UNIVERSITY PRESS INC. 756 Brown DH, Caldwell DK, Caldwell MC. 1966. Observations on the Behavior of Wild and 757 Captive False Killer Whales, With Notes on Associated Behavior of Other Genera of 758 Captive Delphinids. Natural History Museum of Los Angeles County Science Series 759 95: 1-32. 760 Byrne RW, Tanner JE. 2006. Gestural Imitation by a Gorilla: Evidence and Nature of the 761 Capacity. International Journal of Psychology and Psychological Therapy 6: 215-231. 762 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 34 Caldwell MC, Caldwell DK, Siebenaler JB. 1965. Observations on Captive and Wild Atlantic 763 Bottlenosed Dolphins, Tursiops truncatus, in the Northeastern Gulf of Mexico. 764 Contributions in Science 91: 1-10. 765 Call J. 2001. Body Imitation in an Enculturated Orangutan (Pongo pygmaeus). Cybernetics 766 and Systems 32: 97-119. 767 Call J, Carpenter M. 2002. Three Sources of Information in Social Learning. Pages 211-228 768 in Dautenhahn K, Nehaniv CL eds. Imitation in Animals and Artifacts. The MIT Press, 769 Cambridge, MA, US. 770 Call J, Tomasello M. 1995. Use of Social Information in the Problem Solving of Orangutans 771 (Pongo pygmaeus) and Human Children (Homo sapiens). Journal of Comparative 772 Psychology 109: 308-320. 773 Call J, Tomasello M. 1996. The Effect of Humans on the Cognitive Development of Apes. 774 Pages 371-403 in Russon AE , Bard KA and Parker ST eds. Reaching into Thought: 775 The Minds of the Great Apes. Cambridge University Press, New York, NY, US. 776 Connor RC,. Smolker RA, Bejder L. 2006. Synchrony, Social Behaviour and Alliance 777 Affiliation in Indian Ocean Bottlenose Dolphins, Tursiops aduncus. Animal Behaviour 778 72: 1371-1378. 779 Connor RC, Wells RS, Mann J, Read AJ. 2000. The Bottlenose Dophin: Social Relationships 780 in a Fission-Fusion Society. Pages 91-126 in Mann J, Connor RC, Tyack PL, 781 Whitehead H eds. Cetacean Societies: Field Studies of Dolphins and Whales. The 782 University of Chicago Press, Chicago. 783 Custance DM, Whiten A, Bard KA. 1995. Can Young Chimpanzees (Pan troglodytes) Imitate 784 Arbitrary Actions? Hayes and Hayes (1952) Revisited. Behaviour 132: 837-859. 785 Defeyter MA, German TP. 2003. Acquiring an Understanding of Design: Evidence From 786 Children's Insight Problem Solving. Cognition 89: 133-155. 787 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 35 Engleby LK & Powell JR (2019). Detailed observations and mechanisms of mud ring feeding 788 by common bottlenose dolphins (Tursiops truncatus truncatus) in Florida Bay, 789 Florida, USA. Marine Mammal Science 35. 790 Fellner W, Bauer GB, Harley HE. 2006. Cognitive Implications of Synchrony in Dolphins: A 791 Review. Aquatic Mammals 32: 511-516. 792 Grimm D. 2010. Is a Dolphin a Person? Science 327: 1070-1071. 793 Hanus D, Mendes N, Tennie C, & Call J. 2011. Comparing the performances of apes (Gorilla 794 gorilla, Pan troglodytes, Pongo pygmaeus) and human children (Homo sapiens) in the 795 floating peanut task. PloS one 6: e19555. 796 Hayes K J, Hayes C. 1952. Imitation in a Home-Raised Chimpanzee. Quarterly Journal of 797 Experimental Psychology Section B Comparative and Physiological Psychology 45: 798 450-459. 799 Henrich J, Tennie C. 2017. Cultural Evolution in Chimpanzees and Humans. In M. Muller, R. 800 Wrangham, & D. Pilbeam (Eds.), Chimpanzees and Human Evolution (pp. 645–702). 801 Cambridge, MA: Harvard University Press. 802 Herman LM. 1980. Cognitive Characteristics of Dolphins. Pages 363-429 in L. M. Herman 803 ed. Cetacean Behavior: Mechanisms and Functions Wiley Interscience, New York. 804 Herman LM. 2002. Vocal, Social, and Self-Imitation by Bottlenosed Dolphins. Pages 63-108 805 in Dautenhahn K, Nehaniv CL eds. Imitation in Animals and Artifacts. The MIT Press, 806 Cambridge, MA, US. 807 Herzing DL, Pack AA, Delfour F, Starner T, Mason C, Gilliland S, Ramsey C, Kohlsdorf D. 808 (2024). Imitation of computer-generated sounds by wild Atlantic spotted dolphins 809 (Stenella frontalis). Animal Behavior and Cognition 11: 136-166. 810 Heyes CM. 1994. Social learning in animals: categories and mechanisms. Biological 811 Reviews 69: 207-231. 812 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 36 Hopper L M. 2010. ‘Ghost’experiments and the dissection of social learning in humans and 813 animals. Biological Reviews 85: 685-701. 814 Horner V, Whiten A. 2005. Causal Knowledge and Imitation/Emulation Switching in 815 Chimpanzees (Pan troglodytes) and Children (Homo sapiens). Animal Cognition 8: 816 164-181. 817 Horner V, Whiten A. 2007. Learning From Others' Mistakes? Limits on Understanding a 818 Trap-Tube Task by Young Chimpanzees (Pan troglodytes) and Children (Homo 819 sapiens). Journal of Comparative Psychology 121: 12-21. 820 Huang CT, Charman T. 2005. Gradations of Emulation Learning in Infants' Imitation of 821 Actions on Objects. Journal of Experimental Child Psychology 92: 276-302. 822 Kilian A, von Fersen L, Güntürkun O. 2005. Left Hemispheric Advantage for Numerical 823 Abilities in the Bottlenose Dolphin. Behavioural Processes 68: 179-184. 824 Kilian A, Yaman S, von Fersen L, Güntürkün O. 2003. A Bottlenose Dolphin Discriminates 825 Visual Stimuli Differing in Numerosity. Learning & Behavior 31: 133-142. 826 King SL, Sayigh LS, Wells RS, Fellner W & Janik VM 2013 Vocal copying of individually 827 distinctive signature whistles in bottlenose dolphins. Proceedings of the Royal Society: 828 B 280: 20130053. 829 Krützen M, Mann J, Heithaus MR, Connor RC, Bejder L, Sherwin WB. 2005. Cultural 830 Transmission of Tool Use in Bottlenose Dolphins. Proceedings of the National 831 Academy of Sciences of the United States of America 102: 8939-8943. 832 Kuczaj SA II, Gory JD, Xitco MJ Jr. 1998. Using Programs to Solve Problems: Imitation 833 Versus Insight. Behavioral and Brain Sciences 21: 695-696. 834 Kuczaj SA II, Makecha R, Trone M, Paulos RD, Ramos JA. 2006. Role of Peers in Cultural 835 Innovation and Cultural Transmission: Evidence From the Play of Dolphin Calves. 836 International Journal of Comparative Psychology 19: 223-240. 837 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 37 Kuczaj SA II, Walker RT. 2006. How Do Dolphins Solve Problems? Pages 580-601 in 838 Wasserman EA and Zentall TR eds. Comparative Cognition: Experimental 839 Explorations of Animal Intelligence. The MIT Press, Cambridge, MA, US. 840 Kuczaj SA II, Yeater DB. 2006. Dolphin imitation: who, what, when, and why?. Aquatic 841 Mammals 32: 413-422. 842 Kuczaj SA II, Yeater DB, Highfill L. 2012. How selective is social learning in dolphins?. 843 International Journal Of Comparative Psychology 25: 221-236. 844 Laland KN, Janik. VM. 2006. The Animal Cultures Debate. Trends in Ecology & Evolution 845 21: 542-547 846 Maestripieri D, . Whitham JD. 2001. Teaching in Marine Mammals? Anecdotes Versus 847 Science. Behavioral and Brain Sciences 24: 342-343. 848 Mann J, Sargeant B. 2003. Like Mother, Like Calf: The Ontogeny of Foraging Traditions in 849 Wild Indian Ocean Bottlenose Dolphins (Tursiops sp.). Pages 236-266 in Fragaszy 850 DM and Perry S eds. The Biology of Traditions: Models and Evidence. Cambridge 851 University Press, Cambridge. 852 Mann J, Sargeant BL, Minor M. 2007. Calf Inspection of Fish Catches: Opportunities for 853 Oblique Social Learning? Marine Mammal Science 23: 197-202. 854 Mann J, Sargeant BL, Watson-Capps JJ, Gibson QA, Heithaus MR, Connor RC, . Patterson E. 855 2008. Why Do Dolphins Carry Sponges? PLoS ONE 3: e3868. 856 Mann J, Smuts BB. 1999. Behavioral Development in Wild Bottlenose Dolphin Newborns 857 (Tursiops sp.). Behaviour 136: 529-566. 858 Marino L, Connor RC, Fordyce RE, Herman LM, Hof PR, Lefebvre L, Lusseau D, McCowan 859 B, Nimchinsky EA, Pack AA, Rendell L, Reidenberg JS, Reiss D, Uhen MD, Van der 860 Gucht E, Whitehead H. 2007. Cetaceans Have Complex Brains for Complex 861 Cognition. PLoS Biology 5: 966-972. 862 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 38 Mercado E, Uyeyama RK III, Pack AA, Herman LM. 1999. Memory for Action Events in the 863 Bottlenosed Dolphin. Animal Cognition 2: 17-25. 864 Miles HL, Mitchell RW, Harper SE. 1996. Simon Says: The Development of Imitation in an 865 Enculturated Orangutan. Pages 278-299 in Russon AE, Bard KA, Parker ST eds. 866 Reaching into Thought: The Minds of the Great Apes. Cambridge University Press, 867 New York, NY, US. 868 Miles JA, Herzing DL. 2003. Underwater Analysis of the Behavioural Development of Free-869 Ranging Atlantic Spotted Dolphin (Stenella frontalis) Calves (Birth to 4 Years of 870 Age). Aquatic Mammals 29: 363-377. 871 Neadle D, Bandini E, Tennie C. 2020. Testing the individual and social learning abilities of 872 task-naïve captive chimpanzees (Pan troglodytes sp.) in a nut-cracking task. PeerJ 8: 873 e8734. 874 Pace DS. 2000. Fluke-Made Bubble Rings as Toys in Bottlenose Dolphin Calves (Tursiops 875 truncatus). Aquatic Mammals 26: 57-64. 876 Parra GJ. 2007. Observations of an Indo-Pacific Humpback Dolphin Carrying a Sponge: 877 Object Play or Tool Use. Mammalia 71: 147-149. 878 Perelberg A, Schuster R. 2008. Coordinated Breathing in Bottlenose Dolphins (Tursiops 879 truncatus) as Cooperation: Integrating Proximate and Ultimate Explanations. Journal 880 of Comparative Psychology 122: 109-120. 881 Perry SE. 2006. What Cultural Primatology Can Tell Anthropologists About the Evolution of 882 Culture. Annual Review of Anthropology 35: 171-190. 883 Ramirez K. 1999. Animal Training: Successful Animal Management Through Positive 884 Reinforcement. Shedd Aquarium Press, Chicago, IL, US. 885 Reynolds JE, Wells RS, Eide SD. 2000. The Bottlenose Dolphin: Biology and Conservation. 886 University Press of Florida, Gainesville 887 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 39 Richardson MJ, Marsh KL, Schmidt RC. 2005. Effects of Visual and Verbal Interaction on 888 Unintentional Interpersonal Coordination. Journal of Experimental Psychology 889 Human Perception and Performance 31: 62-79. 890 Sargeant BL, Mann J, Berggren P, Krützen M. 2005. Specialization and Development of 891 Beach Hunting, a Rare Foraging Behavior, by Wild Bottlenose Dolphins (Tursiops 892 sp.). Canadian Journal of Zoology 83: 1400-1410. 893 Spence KW. 1937. Experimental Studies of Learning and the Higher Mental Processes in 894 Infrahuman Primates. Psychological Bulletin 34: 806-850. 895 Simões-Lopes PC, Fabián ME, Menegheti JO (1998). Dolphin interactions with the mullet 896 artisanal fishing on southern Brazil: a qualitative and quantitative approach. Revista 897 Brasileira de Zoologia 15: 709-726. 898 Sterelny K. 2009. Peacekeeping in the Culture Wars. Pages 288-304 in Laland KN, Galef BG, 899 Jr. eds. The Question of Animal Culture. Harvard University Press, Cambridge, MA. 900 Tayler CK, Saayman GS. 1973. Imitative Behaviour by Indian Ocean Bottlenose Dolphins 901 (Tursiops aduncus) in Captivity. Behaviour 44: 286-298. 902 Tennie C, Call J, Tomasello M. 2006. Push or Pull: Imitation vs. Emulation in Great Apes and 903 Human Children. Ethology 112: 1159-1169. 904 Tennie C, Call J, Tomasello M. 2009. Ratcheting Up the Ratchet: On the Evolution of 905 Cumulative Culture. Philosophical Transactions of the Royal Society of London B 906 Biological Sciences 364: 2405-2415. 907 Tennie C, Call J, Tomasello M. 2010. Evidence for Emulation in Chimpanzees in Social 908 Settings Using the Floating Peanut Task. PLoS ONE 5: e10544. 909 Tennie C, Call J, Tomasello M. 2012. Untrained chimpanzees (Pan troglodytes 910 schweinfurthii) fail to imitate novel actions. PLoS One 7: e41548. 911 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 40 Tennie C, Premo LS, Braun DR, McPherron SP. 2017. Resetting the null hypothesis: early 912 stone tools and cultural transmission. Current Anthropology 58: 652-72. 913 Tennie C, Bandini E, van Schaik CP & Hopper LM (2020). The zone of latent solutions and 914 its relevance to understanding ape cultures. Biology & Philosophy 35: 1-42. 915 Thorpe WH. 1963. Learning and Instinct in Animals. Methuen, London. 916 Tizzi RA, Castellano, Pace DS. 2001. The Development of Play Behavior in a Bottlenose 917 Dolphin Calf (Tursiops truncatus). Pages 152-157 in Evans PGHPitt-Aiken R, Rogan 918 E eds. Proceedings of the fourteenth Annual Conference of the European Cetacean 919 Society (Cork, Ireland, 2-5 April 2000). 920 Tomasello M. 1990. Cultural Transmission in the Tool Use and Communicatory Signaling of 921 Chimpanzees? Pages 274-311 in S. T. Parker and K. R. Gibson eds. "Language" and 922 Intelligence in Monkeys and Apes: Comparative Developmental Perspectives. 923 Cambridge University Press, New York, NY, US. 924 Tomasello M. 1996. Do Apes Ape? Pages 319-346 in C. M. Heyes and B. G. Galef, Jr. eds. 925 Social Learning in Animals: The Roots of Culture. Academic Press, San Diego, CA, 926 USA. 927 Tomasello M. 1999. The Cultural Origins of Human Cognition. Harvard University Press, 928 Cambridge, MA. 929 Tomasello M, Call J. 1997. Primate Cognition. Oxford University Press, New York, NY, 930 USA. 931 Tomasello M, Call J. 2004. The Role of Humans in the Cognitive Development of Apes 932 Revisited. Animal Cognition 7: 213-215. 933 Tomasello M, Davis-Dasilva M, Camak L, Bard KA. 1987. Observational Learning of Tool-934 Use by Young Chimpanzees. Human Evolution 2: 175-183. 935 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 41 Tomasello M, Kruger AC, Ratner HH. 1993a. Cultural Learning. Behavioral and Brain 936 Sciences 16: 495-511. 937 Tomasello M, Savage-Rumbaugh S, Kruger AC. 1993b. Imitative Learning of Actions on 938 Objects by Children, Chimpanzees, and Enculturated Chimpanzees. Child 939 Development 64: 1688-1705. 940 Torres LG, Read, AJ (2009). Where to catch a fish? The influence of foraging tactics on the 941 ecology of bottlenose dolphins (Tursiops truncatus) in Florida Bay, Florida. Marine 942 Mammal Science 25: 797-815. 943 Want SC, Harris PL. 2001. Learning from Other People's Mistakes: Causal Understanding in 944 Learning to Use a Tool. Child Development 72: 431-443. 945 Whiten A. 2001. Imitation and Cultural Transmission in Apes and Cetaceans. Behavioral and 946 Brain Sciences 24: 359-360. 947 Whiten A, Ham R. 1992. On the Nature and Evolution of Imitation in the Animal Kingdom 948 Reappraisal of a Century of Research. Advances in the Study of Behavior 21: 239-283. 949 Whiten A, Horner V, Litchfield CA, Marshall-Pescini S. 2004. How Do Apes Ape? Learning 950 & Behavior 32: 36-52. 951 Whiten A, van Schaik CP. 2007. The Evolution of Animal ‘Cultures’ and Social Intelligence. 952 Philosophical Transactions of the Royal Society of London B Biological Sciences 362: 953 603-620. 954 Whiten A, Mcguigan N, Marshall-Pescini S, Hopper LM. 2009. Emulation, imitation, over-955 imitation and the scope of culture for child and chimpanzee. Philosophical 956 Transactions of the Royal Society B: Biological Sciences 364: 2417-2428. 957 Whiten A, Goodall J, McGrew WC, Nishida T, Reynolds V, Sugiyama Y, Tutin CEG, 958 Wrangham RW, Boesch C. 1999. Cultures in chimpanzees. Nature 399: 682–685. 959 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint 42 Whiten A, Goodall J, McGrew WC, Nishida T, Reynolds V, Sugiyama Y, Tutin CEG, 960 Wrangham RW, Boesch C. 2001. Charting cultural variation in chimpanzees. 961 Behaviour 138: 1481–1516. 962 Wild S, Allen SJ, Krützen M, King SL, Gerber L, Hoppitt WJ. 2019. Multi-network-based 963 diffusion analysis reveals vertical cultural transmission of sponge tool use within 964 dolphin matrilines. Biology letters 15: 20190227. 965 Wild S, Hoppitt WJ, Allen SJ & Krützen M. 2020. Integrating Genetic, Environmental, and 966 Social Networks to Reveal Transmission Pathways of a Dolphin Foraging 967 Innovation. Current Biology 30: 3024-3030. 968 Wrangham RW, McGrew WC, de Waal FBM & Heltne PG. 1994. Chimpanzee Cultures. 969 Pages xxiii, 424. Harvard University Press, Cambridge, MA, US. 970 Xitco MJ Jr. 1988. Mimicry of Modeled Behaviors by Bottlenose Dolphins. MSc Masther's 971 Thesis, University of Hawaii, Honolulu 124 pp. 972 Zentall TR. 2006. Imitation: Definitions, Evidence, and Mechanisms. Animal Cognition 9: 973 335-353. 974 .CC-BY-NC-ND 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted August 20, 2025. ; https://doi.org/10.1101/2025.08.17.670717doi: bioRxiv preprint