Quantity matching by an orangutan (Pongo abelii)

Transcription

1 DOI /s ORIGINAL PAPER Quantity matching by an orangutan (Pongo abelii) Jennifer Vonk Received: 11 June 2013 / Revised: 8 July 2013 / Accepted: 10 July 2013 Ó Springer-Verlag Berlin Heidelberg 2013 Abstract An adult male orangutan (Pongo abelii) was presented with a series of delayed matching-to-sample (DMTS) tasks in which he was to match images based on (a) the number of individual animals depicted in the photograph (from 1 to 4), (b) the number of abstract shapes presented in the stimulus (from 1 to 4), or (c) the number of dots presented in the stimulus (from 1 to 4, 4 7, or 7 10). The spatial arrangement of the dots and the background color of the stimuli varied, and the size of the dots was manipulated to control for overall ratio of foreground to background. The subject s performance was not affected by these perceptual features, but was affected by the absolute difference and ratio between number of elements in the comparison stimuli. However, the relationship between these variables and his performance was not always linear as predicted by the analog magnitude model. In addition, the subject showed a high degree of transfer to novel numerosities up to ten, indicating that orangutans are capable of estimating quantity for a greater number of items than can presumably be subtilized by humans. Keywords Orangutan Quantity Magnitude Matching-to-sample Introduction Some degree of numerical ability has been established in a wide variety of animal species (for reviews, see Brannon and Roitman 2003; Davis and Perusse 1988; Dehaene J. Vonk (&) Department of Psychology, Oakland University, 2200 N Squirrel Rd., Rochester, MI 48309, USA vonk@oakland.edu 1997). Many distantly related species are capable of relative numerousness judgments. Although primates have been tested most often (Cantlon and Brannon 2006), this list also includes a number of non-primate species such as canids (Baker et al. 2012; West and Young 2002), black bears, (Vonk and Beran 2012), elephants (Perdue et al. 2012), dolphins, (Jaakkola et al. 2005; Kilian et al. 2003), birds (Emmerton 1998; Garland et al. 2012; Pepperberg 2006), amphibians (Uller et al. 2003), and fish (Agrillo et al. 2009; Gomez-Laplaza and Gerlai 2011). From an evolutionary standpoint, it is hardly surprising that many species are capable of estimating differences between quantities as such information provides valuable information when foraging or competing for food. What is of interest from an experimental standpoint is the nature of the underlying representation of quantity that animals utilize. For instance, researchers have evaluated whether animals represent quantity in the form of magnitude judgments, which can be represented perceptually based on mass, density, area etc., or whether they are capable of representing numerosity, which can be distinguished from these perceptual features. Non-humans are capable of estimating the magnitude of different quantities of items and of selecting greater or lesser relative quantities of items, often after extensive training (Anderson et al. 2005, 2007; Beran 2001, 2007; Hauser, Carey and Hauser 2000; Jaakkola et al. 2005; Pepperberg 2006; Thomas et al. 1980; Washburn and Rumbaugh 1991; Vonk and Beran 2012; West and Young 2002). Furthermore, in many prior studies, these discriminations can be performed via a variety of mechanisms independent of true numerical representation, such as perceptual estimation of magnitude, and the use of cues such as surface area and ratio of background to foreground in two-dimensional stimuli. Procedures attempting to

2 control the influence of physical features involve presenting arbitrary stimuli such as dots on a computer screen, controlling for dot size, ratio, and density and having the subjects select the array of items that contains the greater or lesser number of items (Beran 2008; Brannon 2006; Cantlon and Brannon 2006; Vonk and Beran 2012). The results of these studies suggest that numerical estimation in non-humans may be more akin to magnitude estimation than true counting, as performance declines with increased ratio between sets, as predicted by Weber s law (Beran 2008; Brannon and Terrace 2000; Cantlon and Brannon 2006; Jordon and Brannon 2006). However, in some studies, the performance of both monkeys (Beran 2008) and black bears (Vonk and Beran 2012) remained high when amount or area was confounded with number, and even when enumerating subsets within moving arrays, hinting at numerical rather than magnitude representations. However, success under these circumstances tends to be achieved only after a large number of training trials, calling into question the extent to which numerosities are spontaneously represented in non-human species. The majority of studies on animals numerical abilities, and the only data on orangutans, focus on their abilities to make relative numerousness judgments (Anderson et al. 2007; Call 2000; Hanus and Call 2007; Shumaker et al. 2001; Uher and Call 2008). In contrast, the current study focuses on the ability of an adult male orangutan to match stimuli based on absolute numerosities of various twodimensional stimuli. Jordan and colleagues have shown that rhesus macaques can match stimuli representing the numerical value of random shapes or, cross-modally, the number of auditory sounds presented (Jordon and Brannon 2006; Jordan et al. 2008). Nieder et al. (2006) had previously shown that rhesus macaques could match numerosities, but matching paradigms have not commonly been used in other species. Here, a delayed matching-to-sample procedure was used where the sample was no longer visible at the time of test, making it necessary to maintain a representation of quantity in mind, as in Jordan et al. (2008). Orangutans have not previously been tested with regard to their ability to estimate absolute numerosities in a matching paradigm. The ability to match or label stimuli containing more than five elements on the basis of the number of elements in the stimuli might be taken as evidence for absolute numerical representation, assuming that features allowing for magnitude estimation are controlled. The greater number of items is important because items fewer than five can be subtilized (Davis and Perusse 1988; Dehaene 1992; Feigenson et al. 2002). Furthermore, most studies assessing numerosity have relied on either real food rewards as stimuli or abstract shapes such as dots. Prior studies have not tested an animal s ability to quantify the number of individuals in a photograph, and very few have used varying asymmetrical shapes in a two-dimensional stimulus array (see Thomas et al for one example). No such studies have been conducted with orangutans or other apes. These types of stimuli importantly provide natural controls for perceptual cues such as absolute size, magnitude, and density. Following these tests, the orangutan subject of the current study was presented with stimuli containing dot stimuli to provide a comparison of performance with natural and more commonly used, arbitrary, abstract stimuli. Few non-social species have been tested for the ability to estimate numerosities, leaving open the possibility that such an ability arose to track group members rather than to quantify food items (see also Vonk and Beran 2012). Orangutans, unlike many other primates previously tested in such paradigms, do not have the need to track group members. Orangutans have been underrepresented in many cognitive tasks compared to chimpanzees. In addition, the majority of the animals previously tested for their abilities to match, label, or rank numerical stimuli have undergone extensive training in the laboratory (Beran 2001, 2008; Beran et al. 2006; Brannon and Roitman 2003; Jordan et al. 2008). The current orangutan subject had participated in few experimental tasks (Vonk 2002, 2003; Vonk and MacDonald 2004), had never been trained to make any kind of numerical judgments or identity matches, and was not symbol or language trained. Thus, the current study is able to capture relatively untrained representations of quantities. Methods Subject The subject was a 22-year-old adult male Sumatran orangutan (Pongo abelii) named Molek who had participated in a number of touchscreen experiments involving the categorization of natural stimuli (Vonk 2002, 2003; Vonk and MacDonald 2004). None of the prior experiments had involved making numerousness judgements, and only one had involved the presentation of abstract stimuli (Vonk 2003). Molek had previously been trained on the matching-to-sample (MTS) task in a study where he matched items based on sameness of shape or color (Vonk 2003), taxonomic classification (Vonk, in preparation), and social relationship or physical activity (Vonk 2002). In each of the prior 6 experiments that involved matching, Molek participated in sessions. Importantly, Molek had not been trained on an identity matching task but had performed at high levels in the previous conceptual matching experiments. Molek was housed with a group of other orangutans at the Toronto zoo, Toronto, ON, Canada.

3 He participated in these experiments while housed separately off exhibit approximately three afternoons a week. Materials Within each discrimination, 24 novel color photographs or drawings were presented. There were six images within each of four number categories, and the same 24 images were used on each session. Individuals The first discrimination that was presented was the only one in which photographs of animals were used instead of drawings of abstract stimuli. A list of the stimuli appears in Appendix. Animals from various taxa appeared within each numerosity category such that sometimes the incorrect comparison might have appeared physically more similar to the sample. Molek had previously been trained to discriminate and match stimuli based on species and taxonomic categories (Vonk in preparation, Vonk and Mac- Donald 2004) and thus might be expected to make several errors in this experiment when required to attend to the number of elements within the stimulus instead. It was possible that the incorrect comparison stimulus sometimes might have depicted the same or closely related species as the animal(s) in the sample stimulus. An example of stimuli presented on one possible trial appears in Fig. 1 (top). Shapes In this discrimination, Molek was presented with stimuli that included differing numbers of abstract images which were non-symmetrical colored shapes on different colored backgrounds. The components of the stimuli were arranged into varying patterns so that all of the stimuli were perceptually distinct. Once again the number of elements within the stimuli ranged from one to four. An example of the type of stimuli presented on one possible trial appears in Fig. 1 (bottom). Notably, ratio of background to foreground, pattern of the elements, and absolute size of the elements of the stimuli could not be used to aid in discriminating stimuli of varying numerosities in the discriminations Individuals or Shapes. Uncontrolled 1 4 In this discrimination, Molek was presented with black dots on six different colored backgrounds. The background color of the stimuli was manipulated in order to determine whether the orangutan was distracted by perceptual Fig. 1 (top) An example of a possible combination of stimuli for a particular trial in the Individuals discrimination. Here, the sample appears at the top of the figure. On an actual trial, the sample stimulus appeared in the center of the monitor alone and disappeared after being selected by the subject. The two comparison stimuli, depicted at the bottom of the figure, then appeared. (bottom) An example of a possible combination of stimuli for a particular trial in the Shapes discrimination features of the stimuli unrelated to the relevant numerousness values of the stimuli. Thus, each stimulus of the same numerosity appeared on a different colored background, e.g., blue, green, pink, red, yellow, and white. The number of dots ranged from one to four, and the dots within each number value were arranged in the same pattern, except for the number three stimuli in which one of the exemplars (white background) was arranged in a different pattern from the other five exemplars. An example of stimuli presented on one possible trial appears in Fig. 2 (top). Note that here the incorrect comparison stimulus matches the background color of the sample, while the correct comparison stimulus does not. The pattern in which the three dots are displayed differs from the sample to the correct comparison.

4 stimuli with greater numerosities. However, this slight absolute difference in dot size may have provided another cue by which the orangutan may have performed the discriminations. 4 7 In this discrimination, Molek was presented with black dots on different colored backgrounds (yellow, blue, white, green, and red purple) and the number of dots ranged from four to seven. Within the sets of four and six dots, there were four different patterns. Two of the stimuli shared each of the first two arrangements, and only one stimulus was composed of each of the third and fourth arrangements. Within the five-dot set, there were three unique patterns, each shared by two of the stimuli. Within the set of seven dots, there were also four different patterns; however, three of the stimuli shared the same pattern, and the other three exemplars consisted of dots arranged into unique patterns. Once again the size of the dots was varied to control for the ratio of black dots to colored background. An example of stimuli presented on one possible trial appears in Fig. 2 (bottom). Note that here the dot pattern of the five dots in the sample is identical to the dot pattern in the correct stimulus. On other trials, the dot pattern in the sample stimulus would not match that of the correct stimulus Fig. 2 (top) An example of a possible combination of stimuli for a particular trial in the Uncontrolled 1 4 discrimination. Here the background color of the incorrect, but not the correct, comparison stimulus matches that of the sample. (bottom) An example of a possible combination of stimuli for a particular trial in the 4 7 discrimination. Here, the pattern of the dots in the sample and in the correct stimulus is the same Controlled 1 4 In this discrimination, Molek was presented with black dots on different colored backgrounds (same colors as above). The number of dots ranged from one to four, and the dots within each number value were arranged in the same pattern, except for the number three stimuli in which four of the exemplars were arranged in the same pattern and the other two exemplars (blue and white backgrounds) shared a different arrangement. The stimuli also differed from those in the previous discrimination in that the dots were made systematically smaller as the number of elements within the stimuli increased so as to control for the ratio of black dot pattern to background as well as the size of the individual elements of the stimuli. Thus, the proportion of figure to background size was manipulated in all subsequent discriminations such that the dots were smaller in Within this discrimination, the number of dots ranged from seven to ten and the arrangement of dots varied as follows. For the sets of eight, nine, and ten dots, there were three unique pattern arrangements shared by two differently colored background stimuli each. For the seven-dot set, there were four unique patterns. One pattern was shared by three of the stimuli. The other three stimuli within that set were arranged into unique patterns. Once again, the color of backgrounds varied within the stimuli (same colors as for 4 7) and the size of the dots were varied to control for the ratio of black dot to colored background. Procedure The experiment was programmed in Filemaker Pro 3 software and was run on a Macintosh 5300 PowerBook computer. The images were either digitally scanned photographs (Individuals) or were drawn in Deskpaint (All other sets). Each stimulus was approximately 3 by 4 on the 13 Apple touchscreen monitor. The sample stimulus was centered horizontally and vertically on the screen, whereas the two comparison images were separated by approximately 1.5 and were vertically aligned.

5 Testing took place in the holding area of the Toronto Zoo Indo-Malaya pavilion when the orangutan was off exhibit. The monitor was placed against the bars of the subject s holding cage and the subject reached through the mesh holes to touch the screen. The stimuli and the subject s responses could be viewed on the PowerBook placed behind the monitor. In these experiments, the experimenter sat behind the laptop, which was covered by a protective covering, connected to the touchscreen which was pushed right up against the orangutan s enclosure. The images on the touchscreen were mirror-reversed from the images on the laptop, and the experimenter always gazed directly at the midpoint of the screen. The experimenter could not see the orangutan s face or fingers, or the front of the touchscreen, when he made a response so could not react to the correctness or incorrectness of the choice until after the choice was made. The subject was tested at the same time each day and received one to four sessions per day three or four times a week. The subject was not food deprived in any way. Each session consisted of twelve trials. Therefore, each block of five sessions consisted of 60 trials. During a trial, a sample stimulus was presented in the center of the screen and stayed on screen until the subject attended to the sample and touched it; then it disappeared. The two comparison stimuli subsequently appeared on the screen after a brief delay (approximately 3 s). The subject was then required to select, by touching, only one of the two comparison stimuli. If he selected the stimulus that matched the number of elements depicted in the sample, he was given a small food reward (M and Ms or dried fruits and nuts) by hand. After selection of the comparison stimulus and reward, if required, the screen advanced to the next sample stimulus until all twelve trials were completed. There was no time-out for incorrect choices. During the twelve trials within a session, each of the four numerical values was depicted in the sample stimulus three times, with a different exemplar from the category being used for each of the three trials. Each number value was tested against each of the other three numerosities during each session, so that there would be, for instance, three trials where one element appeared in the sample. Different single element stimuli were used as the correct matches (or reinforced comparison stimuli) for these samples on all three of these trials. There were no identity matching trials. On one single element trial, for example, the non-reinforced comparison stimulus depicted two elements. On another single element trial, the nonreinforced comparison stimulus depicted three elements and, on the third trial, the non-reinforced comparison stimulus depicted four elements. Three stimuli within each numerousness value appeared twice within each session, while the other three stimuli within each numerousness value appeared only once per session. The stimuli were sorted randomly before each session. Therefore, the stimuli that appeared twice on some sessions appeared only once on other sessions. If a stimulus appeared twice within a session, it appeared as both a reinforced and a non-reinforced comparison stimulus on different trials, but not as a sample. Therefore, a stimulus that was correct once would be incorrect on its second presentation or vice versa. The stimuli that appeared as samples during a session appeared only once during that session. Side of presentation was counterbalanced so that half of the correct choices appeared on the left and half appeared on the right side of the screen. Order of presentation and pairing of stimuli were randomized. Therefore, each session involved many novel pairings of stimuli within trials. Molek completed twenty-five sessions with the first two discriminations (individuals and shapes), fifteen sessions with the dot patterns 1 4 and 4 7 and eight sessions with the 7 10 dot discrimination. The varying number of sessions reflects the attempt to obtain consistent above chance responding before moving on to novel stimuli. Molek acquired this consistent level of performance more rapidly as he progressed through the discriminations. Molek received the initial sessions with the discriminations in the order described above, although he was not required to reach any predetermined criterion on one type of discrimination before presentation of the next discrimination began. Thus, presentation of subsequent discriminations sometimes commenced before all of the sessions had been completed with the prior discrimination, but not until the first block of five sessions had been completed. Molek was sometimes presented with more than one type of discrimination on the same day in order to help keep the sessions more interesting for him. For each discrimination, binomial tests were conducted on Molek s score for each session of 12 trials. We report the first session for which Molek s performance was above chance for each discrimination. Additionally, we conducted binary logistic regressions of his scores with session number, difference, and ratio as factors. Results Individuals significantly above chance on the first session (two-tailed, n = 12, p = 0.03). Molek s percent correct across five session blocks (60 trials) appears in Fig. 3. A binary logistic regression of his score (correct versus incorrect) on each trial with the absolute numerical difference between

6 Fig. 3 Percent correct scores and standard errors for each block of five sessions (60 trials) in each of the discriminations the number of individuals in the sample and the incorrect choice (absolute difference hereafter), the ratio between the number of individuals in the sample and the incorrect choice (smaller # divided by larger #, ratio hereafter), and session as the independent variables revealed significant effects of absolute difference and ratio (odds ratios = and , p s \ and = 0.01). Note that odds ratios greater than 1.00 indicate that the likelihood of the subject being correct was greater with greater absolute numerical differences between the correct and incorrect stimuli, and greater ratios between them, consistent with Weber s law and predictions of the analog magnitude model. However, the relationship was not perfectly linear as predicted by the model. Table 1 presents Molek s percent correct (and standard errors) as a function of absolute differences between correct and incorrect comparison stimuli for each of the discriminations presented. Shapes significantly above chance on the second session (twotailed, n = 12, p = 0.04). Molek s percent correct across five session blocks (60 trials) appears in Fig. 3. A binomial logistic regression on score with absolute difference, ratio, and session as independent variables revealed significant effects of absolute difference and ratio (odds ratios = and , p s \ 0.001). Again, Molek s performance tended to improve as the absolute difference and ratio between the correct and incorrect stimuli increased (albeit not in a strictly linear fashion). Uncontrolled 1 4 above chance on the fourth session (two-tailed, n =12,p = 0.04). Molek s percent correct across five session blocks (60 trials) appears in Fig. 3. A binary logistic regression was performed on Molek s score on each trial with pattern (same or different), background color (same or different), session, absolute difference, and ratio as variables. There was a significant effect of both absolute difference and ratio on Molek s performance (odds ratios = and , p s = and 0.005, respectively), as well as an effect of session (odds ratio = 1.149, p = 0.005). Again, Molek s performance tended to improve as the absolute difference and ratio between the correct and incorrect stimuli increased (this time in a linear fashion). Performance was also higher in later sessions indicating an effect of learning with this discrimination. Controlled 1 4 significantly above chance on the third session (two-tailed, n = 12, p = 0.04). Molek s percent correct across five session blocks (60 trials) appears in Fig. 3. A binary logistic regression was performed on Molek s score on each trial with session, pattern (same or different), background color (same or different), absolute difference, and ratio as variables. There were significant effects of both absolute difference and session (odds ratios = and 1.116, respectively, both p s = 0.04). Consistent with performance on the previous discriminations, Molek s performance improved as the absolute difference and ratio between the correct and incorrect stimuli increased. 4 7 significantly above chance on the first session (two-tailed, n = 12, p\0.001) indicating that he was able to transfer to

7 Table 1 Percent correct (with standard error in parentheses) for each discrimination, as a function of the absolute numerical difference between the number of elements in the correct and incorrect comparison stimuli Absolute difference Individuals Shapes Uncontrolled (1 4) Controlled (1 4) (0.03) 0.69 (0.04) 0.78 (0.04) 0.80 (0.04) 0.79 (0.03) 0.75 (0.05) (0.04) 0.61 (0.04) 0.79 (0.05) 0.84 (0.05) 0.83 (0.04) 0.92 (0.07) (0.06) 0.96 (0.06) 0.87 (0.07) 0.98 (0.07) 0.70 (0.05) 0.65 (0.10) novel numerosities even though the number of trials on which the sample pattern matched that of the correct choice was substantially reduced. This is true because, in the previous discrimination, numerosities 1, 2, and 4 were always presented in the same pattern, and, in the current discrimination, the patterns of dots varied for all numerical stimuli. Molek s percent correct across five session blocks (60 trials) appears in Fig. 3. A binary logistic regression was performed on Molek s score on each trial with session, pattern (same or different), background color (same or different), absolute difference, and ratio as variables. There were significant effects of absolute difference, ratio, and session (odds ratios = 0.005, 0, and , respectively, all p s = 0.01). Here, the effect of session reflected a decline in performance in the last block of five sessions, rather than an effect of learning significantly above chance by the first session (two-tailed, n = 9 because three trials had to be discarded do to computer error, p = 0.04). Molek s percent correct across five session blocks (60 trials) appears in Fig. 3. A binary logistic regression was performed on Molek s score on each trial with session, pattern (same or different), background color (same or different), absolute difference, and ratio as variables. There were no significant effects (all p s [ 0.05). Discussion Molek quickly reached above chance levels of performance on all discriminations presented to him and showed a high degree of transfer to novel stimulus sets and numerosities. His performance was not influenced by whether the background color of the stimuli or the pattern of the dots in the correct stimulus matched that of the sample (in sets where pattern varied), indicating that he was not distracted or guided by irrelevant (or relevant) perceptual cues, other than the numerosity of the elements in the stimuli (see also Brannon and Terrace 2000; Jordon and Brannon 2006). Furthermore, the ratio of background to foreground in the stimuli did not appear to be a factor in his discrimination performance as this was not a reliable cue in the first two discrimination sets (Individuals and Shapes), and there was no difference in his performance when this feature was controlled or uncontrolled (see Fig. 3). Furthermore, performance did not increase when size of the individual dots could be used as a cue in some of the dot sets. This orangutan s high levels of accuracy in making these matching judgments with numerical stimuli are even more impressive in light of his previous training. He had previously been reinforced for matching stimuli according to biological relatedness in natural category experiments (Vonk in preparation, Vonk and MacDonald 2004). On some trials, making a correct choice in the discrimination involving individuals would be in conflict with that prior training, and yet Molek performed at high levels initially without extensive training. Molek demonstrated effects of learning on only one discrimination Uncontrolled 1 4, which involved the first presentation of dots as stimuli. Thus, for all other discriminations, it appears that accurate matching was not as a function of learning. During some discriminations, performance actually declined across sessions (significantly with discrimination 4 7), an effect which might possibly be explained as boredom with the stimuli. Notably, Molek s performance was perfect on trial one with this stimulus set. Language-trained chimpanzees may also be able to make absolute numerousness judgments spontaneously (Hayes and Nissen 1971; Woodruff and Premack 1981). Extensive training with the use of symbols may have allowed chimpanzee subjects to hone skills not normally called for in a more typical rearing environment. Molek s performance is notable as he was not language or symbol-trained and had experience making only natural category and relational discriminations. Rhesus macaques have been trained to match stimuli up to numerosities of twelve (Jordon and Brannon 2006). However, in the first experiment(s) by these authors, subjects were not presented with a range of distracters they merely discriminated between 2 and 8 and between 3 and 12 in training. Later, they were required to choose matches from a range of numerosities 3 7 and 4 10, but their performance was evaluated to determine whether they estimated larger quantities as being more similar to larger values and vice versa. In experiment 2, they were required

8 to make absolute matches and discrimination was at chance levels for many of the number pairs particularly those farther removed from the trained numerosities of 2, 3, 8, and 9. Sarah, a language-trained chimpanzee could match perceptually dissimilar items on the basis of proportion or numbers 1 4 (Woodruff and Premack 1981). A gray parrot, Alex, has been trained to match stimulus arrays of varying numerosities to symbols representing the cardinal value of the set. Alex can correctly label up to six items independently of mass or contour (Pepperberg 1994; Pepperberg and Gordon 2005). The orangutan, Molek s performance is in line with these findings from other species, suggesting that he can match quantities at least to ten and is likely using a magnitude estimation process. Most of the recent research conducted with nonhuman subjects has indicated a significant and linear effect of ratio and scalar distance on the animals performance (Beran and Rumbaugh 2001; Biro and Matsuzawa 2001; Cantlon and Brannon 2006; Emmerton 1998; Flombaum et al. 2004; Jaakkola et al. 2005; Jordon and Brannon 2006; Smithetal.2003). In addition, rhesus macaques demonstrate more uncertainty and make more errors when making numerosity judgements between stimuli closer to the target value (Beran et al. 2006). Consistent with data reported for other nonhuman species (in pigeons, Emmerton 1998; indolphins, Jaakkola et al in chimpanzees and rhesus macaques, Beran et al. 2006; Beran and Rumbaugh 2001; Biro and Matsuzawa 2001; Cantlon and Brannon 2006; Flombaum et al. 2004; Jordan and Brannon 2006; andin squirrel monkeys and a hamadryas baboon, Smith et al. 2003, see also Olthof et al. 1997), Molek s performance was affected by both the absolute difference and ratio between correct and incorrect stimuli, consistent with the analog magnitude model. The current results are the first reported data on an orangutan s ability to make absolute rather than relative numerical judgments. The data from the individuals and shapes conditions are most compelling because these stimuli did not contain confounds such as dot size present in the later stimuli. Orangutans have previously been shown to outperform chimpanzees in a task requiring them to inhibit the tendency to respond to larger quantities of food rewards (Shumaker et al. 2001). Perhaps orangutans have a more highly developed representation of quantity than other non-human primates tested thus far. This finding would be of interest given that orangutans are less social and would suggest that tracking group members is not critical for the evolution of numerical competence. It is also possible that this particular subject performed especially well on conceptual matching-to-sample tasks because he had not been over-trained on identity matching tasks (see also Vonk et al. 2012). Future studies comparing training methods are necessary to elucidate the role of training in establishing a mechanism for making numerousness judgments in various species. Acknowledgments This research was supported by a Natural Sciences and Engineering Research Council (NSERC) Postgraduate scholarship to the author. The cooperation and support of the staff at the Toronto Zoo was greatly appreciated. Special thanks to Bev Carter, Karyn Tunwell, Bridget Burke-Johnson, Andrea Beatson, Rick Vos, David Partington, Des Macguire, and Jackie Craig, without whose assistance these experiments would not have been possible. I also thank Suzanne MacDonald for providing opportunity and the equipment and Jon Toth for building the protective covering. Thanks to Michael J. Beran and Kelly Jaakola for insightful comments on an earlier draft of this manuscript. Appendix Number of elements Animal depicted 1 Tree boa Meadow vole Walrus Stork Infant rhesus macaque Butterfly fish 2 Badgers Fish Olive baboons Chacma baboon mother and infant Egret chicks Canadian geese 3 Brown Rats Human adult females Orangutans Fish Emus Chimpanzees 4 Burrowing owlets Penguins Orangutans Adult human females Ringtail Lemurs Chimpanzees References Agrillo C, Dadda M, Serena G, Bisazza A (2009) Use of number by fish. PLoS ONE 4:e4786. doi: /journal.pone Anderson US, Stoinski TS, Bloomsmith MA, Marr MJ, Smith AD, Maple TL (2005) Relative numerousness judgment and summation in young and old western lowland gorillas. J Comp Psychol 119:

9 Anderson US, Stoinski TS, Bloomsmith MA, Maple TL (2007) Relative numerousness judgment and summation in young, middle-aged, and older adult orangutans (Pongo pygmaeus abelii and Pongo pygmaeus pygmaeus). J Comp Psychol 121:1 11. doi: / Baker JM, Morath J, Rodzon KS, Jordan KE (2012) A shared system of representation governing quantity discrimination in canids. Frontiers Psychol 3. doi: /fpsyg Beran MJ (2001) Summation and numerousness judgments of sequentially presented sets of items by chimpanzees (Pan troglodytes). J Comp Psychol 115: Beran MJ (2007) Rhesus monkeys (Macaca mulatta) enumerate sequentially presented sets of items using analog numerical representations. J Exp Psychol Anim Behav Process 33:42 54 Beran MJ (2008) Monkeys (Macaca mulatta and Cebus apella) track, enumerate and compare multiple sets of moving items. J Exp Psychol Anim Behav Process 34:63 74 Beran MJ, Rumbaugh DM (2001) Constructive enumeration by chimpanzees (Pan troglodytes) on a computerized task. Anim Cog 4:81 89 Beran MJ, Smith JD, Redford JS, Washburn DA (2006) Rhesus macaques (Macaca mulatta) monitor uncertainty during numerosity judgments. J Exp Psychol Anim Behav Process 32: Biro D, Matsuzawa T (2001) Chimpanzee numerical competence: cardinal and ordinal skills. In: Matsuzawa T (ed) Primate origins of human cognition and behavior. Springer, New York, pp Brannon EM (2006) The representation of numerical magnitude. Curr Opinion Neurobiol 16: Brannon EM, Roitman JD (2003) Nonverbal representations of time and number in animals and human infants. CRC Press, Boca Raton, FL, pp doi: / ch6 Brannon EM, Terrace HS (2000) Representation of the numerosities 1 9 by rhesus macaques (Macaca mulatta). J Exp Psychol Anim Behav Process 26:31 49 Call J (2000) Estimating and operating on discrete quantities in orangutans (Pongo pygmaeus). J Comp Psychol 114: Cantlon JF, Brannon EM (2006) Shared system for ordering small and large numbers in monkeys and humans. Psycho Sci 17: Davis H, Perusse R (1988) Numerical competence in animals: definitional issues, current evidence and a new research agenda. Behav Brain Sci 11: Dehaene S (1992) Varieties of numerical abilities. Cognition 44:1 42 Dehaene S (1997) The number sense: how the mind creates mathematics. Oxford University Press, New York Emmerton J (1998) Numerosity differences and effects of stimulus density on pigeons discrimination performance. Anim Learn Behav 26: Feigenson L, Carey S, Hauser M (2002) The representations underlying infants choice of more: object files versus analog magnitudes. Psychol Sci 13: Flombaum JI, Junge JA, Hauser MD (2004) Rhesus macaques (Macaca mulatta) spontaneously compute addition operations over large numbers. Cognition 97: Garland A, Low J, Burns KC (2012) Large quantity discrimination by north island robins (Petroica longipes). Anim Cog 15: doi: /s Gomez-Laplaza LM, Gerlai R (2011) Can angelfish (Pterophyllum scalare) count? Discrimination among different shoal sizes follows Weber s law. Anim Cogn 14:1 9 Hanus D, Call J (2007) Discrete quantity judgments in the great apes (Pan paniscus, Pan troglodytes, Gorilla gorilla, Pongo pygmaeus): the effect of presenting whole sets versus item-by-item. J Comp Psychol 121: doi: / Hauser MD, Carey S, Hauser LB (2000) Spontaneous number representation in semi-free-ranging rhesus monkeys. Proc R Soc Lond Biol Sci 267: Hayes KJ, Nissen CH (1971) Higher mental function of a home-raised chimpanzee. In: Schrier AM, Stollnitz F (eds) Behavior of nonhuman primates. Academic Press, New York, pp Jaakkola K, Fellner W, Erb L, Rodriguez M, Guarino E (2005) Understanding of the concept Numerically less by Bottlenose Dolphins (Tursiops truncatus). J Comp Psychol 119: Jordan KE, MacLean EL, Brannon EM (2008) Monkeys match and tally quantities across senses. Cognition 108: doi: /j.cognition Jordon KE, Brannon EM (2006) Weber s law influences numerical representations in rhesus macaques (Macaca mulatta). Anim Cog 9: Kilian A, Yaman S, von Fersen L, Gunturkun O (2003) A bottlenose dolphin discriminates visual stimuli differing in numerosity. Learn Behav 31: Nieder A, Diester I, Tudusciuc O (2006) Temporal and spatial enumeration processes in the primate parietal cortex. Science 313: doi: /science Olthof A, Iden CM, Roberts WA (1997) Judgments of ordinality and summation of number symbols by squirrel monkeys (Saimiri sciureus). J Exp Psychol Anim Behav Process 23: Pepperberg IM (1994) Numerical competence in an African gray parrot (Psittacus Erithacus). J Comp Psychol 108:36 44 Pepperberg IM (2006) Ordinality and inferential abilities of a grey parrot (Psittacus Erithacus). J Comp Psychol 120: Pepperberg IM, Gordon JD (2005) Number comprehension by a grey parrot (Psittacus erithacus), including a zero-like concept. J Comp Psychol 119: Perdue BM, Talbot CF, Stone AM, Beran MJ (2012) Putting the elephant back in the herd: elephant relative quantity judgments match those of other species. Anim Cog 15: doi: /s y Shumaker RW, Palkovich AM, Beck BB, Guagnano GA, Morowitz H (2001) Spontaneous use of magnitude discrimination and ordination by the orangutan (Pongo pygmaeus). J Comp Psychol 115: Smith BR, Piel AK, Candland DK (2003) Numerity of a socially housed hamadryas baboon (Papio hamadryas) and a socially housed squirrel monkey (Saimiri sciureus). J Comp Psychol 117: Thomas RK, Fowlkes D, Vickery JD (1980) Conceptual numerousness judgements by squirrel monkeys. Am J Psycho 93: Uher J, Call J (2008) How the great apes (Pan troglodytes, Pongo pygmaeus, Pan paniscus, Gorilla gorilla) perform on the reversed reward contingency task II: transfer to new quantities, long-term retention, and the impact of quantity ratios. J Comp Psychol 122: doi: / Uller C, Jaeger R, Guidry G, Martin C (2003) Salamanders (Plethodon cinereus) go for more: rudiments of number in an amphibian. Anim Cogn 6: Vonk J (2002) Can Orangutans (Pongo abelii) and Gorillas (Gorilla gorilla gorilla) acquire concepts for social relationships? Int J Comp Cog 15: Vonk J (2003) Gorilla (Gorilla gorilla gorilla) and Orangutan (Pongo abelii) understanding of first and second order relations. Anim Cog 6:77 86 Vonk J, Beran MJ (2012) Bears count too: quantity estimation and comparison in black bears, ursus americanus. Anim Behav 84: doi: /j.anbehav Vonk J, MacDonald SE (2004) Levels of abstraction in Orangutan (Pongo abelii) categorization. J Comp Psychol 118:3 13

10 Vonk J, Jett SE, Mosteller KW (2012) Concept formation in American black bears (Ursus americanus). Anim Behav 84: Washburn DA, Rumbaugh DM (1991) Ordinal judgements of numerical symbols by macaques (Macaca mulatta). Psychol Sci 2: West RE, Young RJ (2002) Do domestic dogs show any evidence of being able to count? Anim Cog 5: Woodruff G, Premack D (1981) Primitive mathematical concepts in the chimpanzee: proportionality and numerosity. Nature 293: