Psychon Bull Rev DOI 10.3758/s13423-013-0573-2
BRIEF REPORT
Do rhesus monkeys (Macaca mulatta) perceive the Zöllner illusion? Christian Agrillo & Audrey E. Parrish & Michael J. Beran
# Psychonomic Society, Inc. 2014
Abstract A long-standing debate surrounds the issue of whether human and nonhuman animals share the same perceptual mechanisms. In humans, the Zöllner illusion occurs when two parallel lines appear to be convergent when oblique crosshatching lines are superimposed. Although one baboon study suggests that they too might perceive this illusion, the results of that study were unclear, whereas two recent studies suggest that birds see this illusion in the opposite direction from humans. It is currently unclear whether these mixed results are an artifact of the experimental design or reflect a peculiarity of birds’ visual system or, instead, a wider phenomenon shared among nonhuman mammals. Here, we trained 6 monkeys to select the narrower of two gaps at the end of two convergent lines. Three different conditions were set up: control (no crosshatches), perpendicular (crosshatches not inducing the illusion), and Zöllner (crosshatches inducing the illusion in humans). During training, the degrees of convergence between the two lines ranged from 15° to 12°. Monkeys that reached the training criterion were tested with more difficult discriminations (11°–1°), including probe trials with parallel lines (0°). The results showed that monkeys perceived the Zöllner illusion in the same direction as humans. Comparison of these data with the data from bird studies points toward the existence of different orientation-tuned mechanisms between primate and nonprimate species.
C. Agrillo (*) Department of General Psychology, University of Padova, Via Venezia 8, 35131 Padova, Italy e-mail: [email protected] A. E. Parrish : M. J. Beran Language Research Center, Georgia State University, Atlanta, USA A. E. Parrish Department of Psychology, Georgia State University, Atlanta, USA
Keywords Visual illusion . Zöllner illusion . Primates . Macaca mulatta . Comparative perception
Introduction The study of visual perception represents one of the research areas that have made the most progress in cognitive sciences during the last 30 years. This is partly due to the increased use of animal models as a tool for understanding the neurocognitive systems that are involved in processing behaviorally relevant information from the retina (e.g., Baroncelli, Braschi, & Maffei, 2013; Cook, Qadri, Kieres, & CommonsMiller, 2012; Dadda, Domenichini, Piffer, Argenton, & Bisazza, 2010). However, despite this growing interest in nonhuman animal visual systems, it is still unclear whether other vertebrates share with humans similar core systems to analyze shape, colors, depth, and the orientation of objects, and it remains unknown to what extent the Gestalt principles (Koffka, 1955; Wertheimer, 1938) described in humans can be generalized to nonhuman animals (Nieder, 2002). Research into visual illusions represents a valuable tool for understanding the mechanisms underlying visual perception, and it can reveal the hidden constraints of the perceptual system. For a long time, the study of visual illusions remained an exclusive area of research for neuroscientists studying humans. But in the last few decades, attention has shifted to include studies with other mammals (e.g., Beran, 2006; Bravo, Blake, & Morrison 1988) and birds (e.g., Nakamura, Watanabe, & Fujita, 2008; Pepperberg, Vicinay, & Cavanagh, 2008). Two animal models, in particular, are frequently adopted in this research field: rhesus monkeys (Macaca mulatta; e.g., Beran & Parrish, 2013; Fujta, 1997; Huang, MacEvoy, & Paradiso, 2002; Zivotofsky, Goldberg, & Powell, 2005) and pigeons (Columba livia; e.g., Cook et al., 2012; Fujita, Blough, & Blough, 1991; Nakamura et al., 2008).
Psychon Bull Rev
One important visual illusion is called the Zöllner illusion. In its classical configuration, two parallel lines appear to be convergent when oblique crosshatching lines are superimposed (see Fig. 1b, column 3). The explanation of this illusion is still uncertain. Several authors believe that the perceptual expansion of acute angles is the main reason underlying this phenomenon (e.g., Kitaoka & Ishihara, 2000; Oyama, 1975; Wallace, 1969). The acute angles formed by the intersections of the lines with the crosshatches would be perceptually enlarged; such expansion would cause a slight perceived change in the orientation of the whole line such that two parallel lines appear to converge. In other words, the illusion would be the result of the sum of the effects of the individual angles (White, 1972). Alternative explanations invoke the role of global characteristics of stimulus configuration (Parlangeli & Roncato, 1995) and viewing distance (e.g., Wallace, 1969). In order to investigate the perception of the Zöllner illusion in nonhuman animals, Watanabe, Nakamura, and Fujita (2011) trained pigeons to peck at the narrower (or wider) of two gaps at the end of a pair of convergent lines. In the test phase, the pigeons were presented with parallel lines with crosshatches, a presentation that typically induces the Zöllner illusion in humans. Watanabe et al. (2011) found that pigeons perceived a form of the illusion, selecting one gap more than would be expected by chance. Interestingly, however, the direction was opposite to that of humans: Pigeons
selected as the narrower gap the one perceived as wider by human subjects. Thus, the stimuli that led to the Zöllner illusion among humans actually improved performance among the pigeons. Watanabe et al. (2011) hypothesized that pigeons underestimated the acute angles between each main line and its crosshatches. The same (reversed) illusion was reported in another bird species as well, the bantam (Gallus gallus domesticus; Watanabe, Nakamura, & Fujita, 2013). Both of these studies raise the intriguing question as to whether the perception of a reversed Zöllner illusion reflects a peculiarity of birds’ visual system or, instead, represents a more pervasive phenomenon of vertebrates’ visual system that somehow is not reflected in the perception of humans. The only attempt to study the Zöllner illusion in mammals was made by Benhar and Samuel (1982). Two baboons were tested in an oddity task in which they were trained to select the odd stimulus in a triad. When presented with two parallel lines (stimulus 1), two convergent lines (stimulus 2), and two “Zöllner” lines (stimulus 3), subjects significantly selected parallel lines as the odd stimulus, presumably showing that the baboons perceived the Zöllner lines as convergent. However, the procedure used in that study did not allow for the establishment of the direction of the illusion in baboons, and so the possibility remained that nonhuman primates might also perceive a reversed Zöllner illusion when an experimental setup similar to those in the recent studies with humans and birds is used.
Fig. 1 Example of testing stimuli. Monkeys were trained to select the narrower gap between the lines’ edges. In the training phase (a), convergences of the main lines ranged from 15° to 12°. In the test phase (b), more difficult discriminations were presented (11°–1°), including parallel
lines. For each phase, three different conditions were presented: control (no crosshatch), perpendicular (90° crosshatches), and Zöllner (30° crosshatches) conditions
Psychon Bull Rev
In the present study, we investigated the perception of the Zöllner illusion in rhesus monkeys. Previous studies showed that rhesus monkeys perceived the Ponzo illusion (Fujta, 1997), the Dunker illusion (Zivotofsky et al., 2005), and the brightness illusion (Huang et al., 2002) in a qualitative fashion that resembles the illusion as seen in human perception, thus making macaques proper candidates for assessing the directionality of any Zöllner illusion in nonhuman animals. In particular, we assessed (1) whether rhesus monkeys perceived this illusory convergence and (2) the direction of the illusion if it occurred. To achieve these goals, monkeys were trained to select the narrower of the two gaps at the end of a pair of convergent lines. Once they reached the learning criterion with easy discriminations, monkeys underwent harder trials, including parallel lines. Three different conditions were set up: control (lines without crosshatches), perpendicular (90° crosshatches to the lines, which result in no illusion for humans), and Zöllner (30° crosshatches to the lines, a presentation that generated the illusory pattern in humans; Wallace & Moulden, 1973; Watanabe et al., 2011). If monkeys perceived the illusory convergence at all, they were expected to perceive parallel lines as converging only in the Zöllner condition, due to the nature of the crosshatches. If that occurred, the direction of their choice would also reveal whether they perceived a humanlike or birdlike (reversed) illusion.
Method Subjects We tested 6 male rhesus macaques from the Language Research Center in Atlanta, GA, including Hank (age 29), Gale (age 29), Luke (age 13), Lou (age 19), Murph (age 19), and Obi (age 9). All monkeys were individually housed with constant visual and auditory access to conspecifics and were given weekly outdoor access with another monkey. Monkeys had constant access to their computerized testing apparatus when they were in the home cage, and thus, they could freely engage tasks to earn food pellets throughout the day. They were fed primate chow, fruits, and vegetables once a day no matter how many experimental trials they completed, and they had 24-h access to water. No animals were food or water deprived for testing purposes. The present study complied with protocols approved by the Georgia State University IACUC and was in full accordance with the USDA Animal Welfare Act and the “Guidelines for the Use of Laboratory Animals.” Materials Testing was conducted via the Language Research Center’s Computerized Test System, consisting of a personal computer,
digital joystick, 17-in. color monitor, and food pellet dispenser (Evans, Beran, Chan, Klein, & Menzel, 2008; Richardson, Washburn, Hopkins, Savage-Rumbaugh, & Rumbaugh, 1990). Monkeys manipulated a digital joystick that generated isomorphic movements of a cursor on the attached screen. The computer program was written using Microsoft Visual Basic 6.0. Food rewards were 94-mg Bio-Serv food pellets that were dispensed contingent upon correct responses. All monkeys had extensive experience with computerized testing, including perceptual illusion experiments (e.g., visually nested stimuli; Beran & Parrish, 2013). Stimuli Stimuli consisted of two black lines that were center justified on the screen. Lines were 89.1 mm long and 0.79 mm thick. Each pattern (the two lines together) was at an angle of approximately 45° with respect to the horizontal axis of the computer screen. Crosshatches intersected the main lines at the crosshatches’ midpoint, were 11.9 mm long and 0.79 mm thick, and were equidistant from each other. We used 10 crosshatches per main line, as was the procedure in the two most recent Zöllner studies (Watanabe et al., 2011, 2013). Crosshatches in the perpendicular condition were placed at a 90° angle to the main lines and a 30° angle to main lines in the Zöllner condition (also consistent with Watanabe et al., 2011, 2013). Therefore, perpendicular and Zöllner conditions presented the same global features (two identical lines and 10 identical crosshatches equally spaced in the two sets). The only difference was the orientation degree of the crosshatches with respect to the main lines. Thus, the Zöllner condition should have exclusively elicited the illusory pattern. However, should the presence of crosshatches have induced any other bias not related to the illusion, this should have occurred in both of these conditions. We presented 31° of convergence (ranging from +15° to −15°), depending upon the phase (see below). Positive and negative degrees varied the side of the smaller gap (right or left) so that each angle was presented equally on both sides. In addition, to prevent the possibility that monkeys could have used other spatial strategies, we randomized the location of the smaller gap so that it appeared in the top and bottom corners of the screen half of the time. We also randomized the side on which the smaller gap appeared, such that it was located on the right and left sides of the screen half of the time. In short, the narrower gap appeared 25 % of the time in the corner “left-above,” 25 % of the time in the corner “leftbelow,” 25 % of the time in the corner “right-above,” and 25 % of the time in the corner “right-below,” thus preventing any possibility of using spatial cues. Normal viewing distance ranged from 30.5 to 40.6 cm, a distance commonly used to investigate visual perception in rhesus monkeys (e.g., Agrillo et al., in preparation; Beran & Parrish, 2013).
Psychon Bull Rev
General procedure
Testing
Each trial presented the monkeys with a cursor that they controlled, the two main lines of variable convergence and crosshatching, and two choice icons (a black leftwardpointing arrow and a black rightward-pointing arrow). The distance between the choice icons was 11 cm. The distance between stimulus and choice icons was 3.5 cm. The arrows were bottom justified on their appropriate side (i.e., the left arrow was left-bottom justified and the right arrow was rightbottom justified). The monkeys controlled a small round cursor that was centered between the choice icons via the joystick. Monkeys moved that cursor into contact with one of the choice icons to indicate which side they perceived to be the side where the main lines converged (i.e., which side had the smaller gap). Selecting the correct choice icon led to one pellet being delivered, whereas selecting the incorrect choice icon led to a 20 s timeout during which the computer screen remained blank. The intertrial interval was 1 s. There was no requirement for a speeded response from the monkeys, since the stimuli would stay on the screen until a monkey made a response. Monkeys were tested in 2 to 4-h sessions at a maximum of once per day. We presented three conditions, including a control condition with no crosshatches, a perpendicular condition with 90° crosshatches, and a Zöllner condition with 30° crosshatches (see Fig. 1). Conditions were randomly presented within a session, with equivalent numbers of each condition per session.
Twenty-three new angles were introduced in the test phase for all three conditions, along with the training stimuli (±15° through ±12), including +11°, +10°, +9°, +8°, +7°, +6°, +5°, +4°, +3°, +2°, +1°, 0°, −11°, −10°, −9°, −8°, −7°, −6°, −5°, −4°, −3°, −2°, and −1°. We included the critical probe trials of the parallel lines (0°) for all three conditions (see Fig. 1b). Again, these probe trials in the Zöllner condition should induce the illusion that parallel lines converge at one end due to the nature of the angled crosshatches. Responses to these probe trials were not reinforced or punished for either response from the monkeys. They simply led directly to the intertrial interval and then the next trial presentation. In order to have a sufficient number of trials to allow us to generate smooth choice curves with the large range of convergence angles presented in the three conditions, the first 8,000 test trials were used in data analysis; monkeys could complete up to 2,000 test trials per session.
Training Eight angles were included in the preliminary training phase for all three conditions, including +15°, +14°, +13°, +12°, −12°, −13°, −14°, and −15°. The criterion for moving to the test phase was 75 % correct choices on each condition in each of two consecutive sessions of at least 250 trials in each condition. If a monkey failed to reach criterion within five sessions, we introduced six larger angles (+30°, +27°, +25°, +22°, +20°, +18°, −18°, −20°, −22°, −25°, −27°, and −30°), a condition called “pretraining 1.” The criterion was the same as above. If a monkey still failed to reach criterion, we introduced only the control condition, in which the bottom main line was parallel with the horizontal axis of the computer screen and, thus, had no pitch (pretraining 2). In that case, the angles were ±15° through ±12°, as above. Criterion was 75 % correct choices in one session of at least 250 trials. Next, they advanced back to the first training phase with angles of ±15° through ±12° for all conditions. If monkeys performed eight consecutive sessions without exhibiting any increase in performance, they were discontinued as subjects in this experiment. Figure 2 summarizes the procedure adopted.
Results Three out of 6 monkeys never reached 75 % correct choices on each condition in two consecutive sessions of the training phase and, hence, did not start the test phase. Hank was discontinued as a subject after pretraining 2, performing a total 7,458 trials, with an average accuracy in the last two sessions of .479. Gale was discontinued as a subject in the pretraining 2 phase, performing a total of 5,041 trials, with an average accuracy in the last two sessions of .442. Obi was discontinued as a subject in the training phase, performing 12,205 trials, with an average accuracy in the last two sessions of .529. Murph, Luke, and Lou reached the learning criterion. During the training phase, Murph performed a total of 5,891 trials, reaching an average accuracy in the last two sessions of .846, .833, and .840 for the control, perpendicular, and Zöllner conditions, respectively. Luke performed a total of 9,175 trials, reaching an accuracy of .875, .830, and .843 for the control, perpendicular and Zöllner conditions, and Lou performed 5,113 trials, with an accuracy of .912, .982, and .908 for the control, perpendicular, and Zöllner conditions. In the test phase, separate two-tailed binomial tests were conducted for the parallel lines presented in each condition. Given the high number of trials per condition, normal approximation to the binomial test (z-scores) was applied. In the Zöllner condition, the performance of all subjects differed from chance levels (Murph N = 156, df = 1, Z = 8.647, p < .05; Lou, N = 196, df = 1, Z = 6.857, p < .05; Luke, N = 107, df = 1, Z = 2.998, p < .05). In all cases, the monkeys’ bias was in the same direction as that reported for humans (Fig. 3).
Psychon Bull Rev
Fig. 2 Schematic representation of the procedure. All monkeys initially started the training phase (1). Subjects reaching the learning criterion began the test phase, which included more difficult discriminations and parallel lines (2a). Those monkeys who did not reach the criterion were, instead, presented with easier discriminations (2b). If they reached the learning criterion, they started again the training phase (1); otherwise,
they were presented with an easier pattern of lines in which the main line was parallel to the horizontal axis (2c). Monkeys were discontinued as subjects if their performance did not show any signs of increasing in proficiency after eight consecutive sessions in the training phase or the pretraining phase
In contrast, no significant choice bias from chance was observed in the control (Murph, N = 193, df = 1, Z = 1.223, p > .05; Lou, N = 216, df = 1, Z = 1.499, p > .05) and perpendicular (Murph, N = 160, df = 1, Z = 0.791, p > .05; Lou, N = 206, df = 1, Z = 0.418, p > .05; Luke, N = 129, df = 1, Z = 1.321, p > .05) conditions, with the exception of Luke’s
performance in the control condition (N = 98, df = 1, Z = 3.302, p < .05). Figure 4 shows the percentage of trials correctly completed as a function of the degree of convergence. For those trials with truly convergent lines, a repeated measures ANOVA with degree of convergence (1–11) and condition (control/perpendicular/Zöllner) as within-subjects factors found main effects of degree of convergence, F(10, 20) = 20.36, p < .001, ηp2 = .816, and condition, F(2, 4) = 8.87, p = .034, ηp2 = .911, indicating that the monkeys’ accuracy improved as the degree of convergence was increased and that their performance varied as a function of condition. Post hoc analyses (LSD) revealed a significant difference between the Zöllner condition and the perpendicular condition (p = .043). No difference was found between the control and perpendicular conditions (p = .422) and between the control and Zöllner conditions (p = .093). A significant interaction also was found, F(20, 40) = 4.74, p < .001, ηp2 = .703, meaning that the slopes of the three conditions were different as a function of the degree of convergence. However, the lack of a significant difference between the control and the Zöllner conditions might be partially due to a potential ceiling effect, considering that easy discriminations were also included in the test phase and that, on these trials,
Fig. 3 Results of test phase, parallel lines: Percentage of trials on which the monkeys selected the end that appears to converge under the illusion as seen by humans. In the Zöllner condition, all monkeys selected the same direction of convergence as that perceived by humans at levels greater than chance
Psychon Bull Rev
Fig. 4 Results of the test phase, convergent lines: Percentage of correct trials plotted against the different degrees of convergence. Overall performance decreases as the degree of convergence decreases. The slopes of the three conditions significantly differ from each other, with an overall
higher performance in the Zöllner condition in the presence of more difficult discriminations (1°–3°). Bars in the average performance represent the standard error of the mean
there was not much need for a facilitation that could occur through the Zöllner illusion. To control for this aspect, we subsequently analyzed whether performance varied among the conditions for only the three smallest degrees of true convergence (1°–3°). Now, there was a main effect of condition, F(2, 4) = 44.42, p < .001, ηp2 = .957. There was no main effect of degree of convergence, F(2, 4) = 6.12, p = .060, ηp2 = .755, and there was no interaction, F(4, 8) = 0.30, p = .871, ηp2 = .130. Post hoc analyses (LSD) showed that better performance occurred in the Zöllner condition, as compared with the two other conditions: Zöllner versus control, p = .010; Zöllner versus perpendicular, p = .025; control versus perpendicular, p = .890.
with parallel lines. The present results indicated that monkeys perceived the Zöllner illusion in the same direction as humans. When there was no correct answer because the lines were parallel and there were 30° crosshatches as are typically used in presenting the Zöllner illusion, all monkeys were biased to indicate that the lines converged toward the end for which humans also perceive parallel lines as being more convergent. When we analyzed the performance in the presence of convergent lines, accuracy decreased with decreases in the degree of convergence. However, a different pattern was found as a function of stimulus type. In the Zöllner condition, the percentage of correct choices was less negatively affected by the degree of convergence. In particular, the monkeys’ overall accuracy was higher than that reported in the perpendicular condition, a condition characterized by the same overall stimulus features. Since the degree of convergence of crosshatches is the only difference between the two conditions, the different performance in the Zöllner condition can be ascribed only to this variable: 30° crosshatches led to an increased perception of convergence between the lines, thus increasing the monkeys’ ability to select the narrower gap. Not only do these results together indicate that rhesus monkeys perceive the
Discussion The present study assessed whether rhesus monkeys perceived the Zöllner illusion. Monkeys were trained with convergent lines having no crosshatches (control condition), 90° crosshatches (perpendicular condition), and 30° crosshatches (Zöllner condition) and then were observed in probe trials
Psychon Bull Rev
Zöllner illusion, but also this study confirms that they too show the illusion in the same direction as humans, rather than the reverse pattern that normally is reported in birds (Watanabe et al., 2011, 2013). It is interesting to note that 3 subjects out of 6 did not reach the learning criterion. Unfortunately, we can only speculate about the reasons underlying their lack of discrimination, since they otherwise have performed comparably on other perceptual and cognitive tasks to the macaques that were successful in this study. One possibility is that these subjects were not particularly skillful in visual discrimination or were not motivated to accomplish the task. Two of these monkeys were the oldest individuals of the sample, a factor that might be partially related to lack of motivation or poorer visual acuity. Also, we must acknowledge the possibility that previous computerized tasks might have partially affected the performance: All monkeys had previous experience with numerous cognitive-perceptual computerized tasks. Those past experiences might have generated interference in their ability to learn the basic parameters of this task. One might also argue that the lack of learning to perform this discrimination might be due to specific difficulties in discriminating visual stimuli with the particular characteristics that are relevant to assessing this visual line illusion, such as converging lines with crosshatches. However, we do not feel that this is the case, since similar (low) performances were found for the control, perpendicular, and Zöllner conditions in the last sessions of their training phase. Previous studies have shown that humans and rhesus monkeys share similar Gestalt principles in their perceptual experiences (Fujta, 1997; Tudusciuc & Nieder, 2010; Zivotofsky et al., 2005). Our study aligns with this literature, since the Zöllner illusion emerges due to the global processing and subsequent interaction of all elements within the visual array, including main lines and crosshatches. These and similar findings indicate the value of using rhesus monkeys as a model to investigate the mechanisms underlying objects’ orientation too. For instance, the cortical areas involved in the expansion of acute angles in Zöllner patterns are partly unknown. It has been hypothesized (Kitaoka & Ishihara, 2000) that there is an involvement of areas V1, V2, and V4, as well as higher areas such as the inferotemporal cortex (TE). Rhesus monkeys are commonly used to study neuroanatomical correlates of perceptual and cognitive functions in primates (Desimone & Gross, 1979; Nieder & Merten, 2007), and their visuotopic organization of visual areas from V1 through V4 and of area TE have been extensively mapped (Boussaoud, Desimone, & Ungerleider, 1991). In this sense, future macaque brain imaging studies could help us to shed light into the neural networks involved in the Zöllner illusion. This study also raises the intriguing question about why humans and rhesus monkeys perceive the Zöllner illusion differently from pigeons and bantams. With respect of this
topic, Watanabe et al. (2011) referred to the contrast and assimilation hypothesis advanced by Nakamura et al. (2008). According to this hypothesis, birds would experience both “assimilation” (the phenomenon according to which the perception of two figures tends to assimilate and one figure looks similar to the other in its perceptual features) and “contrast” (the opposite phenomenon according to which the perceptual differences between two figures are enhanced) effects. Birds, however, would be less sensitive to contrast effects, as compared with humans (Watanabe et al., 2011) and, presumably, rhesus monkeys. As a consequence, they would underestimate the acute angles between lines and crosshatches in the Zöllner pattern as a result of assimilation of orientation of two lines, leading to a reversed Zöllner illusion. In line with these findings, previous studies showed that pigeons display different perceptual mechanisms, as compared with those described in primates: Pigeons do not perceive some visual illusions in the same way as humans (e.g., Ebbinghaus-Titchener and Müller-Lyer illusions; Nakamura, Fujita, Ushitani, & Miyatat, 2006; Nakamura et al., 2008), and there is also debate as to whether they can exhibit amodal completion (Fujita 2004; Nagasaka, Lazareva, & Wasserman, 2007). The Zöllner illusion in our species is affected by global characteristics of stimulus configuration (e.g., Parlangeli & Roncato, 1995), and the possibility remains that monkeys might be more attentive to global properties of the images than are pigeons/chickens. Geometric illusions, like the Zöllner illusion, are critically dependent upon the simultaneous processing of the entire visual array, including the illusory inducing context and the target stimulus (e.g., Parron & Fagot, 2007; Roberts, Harris, & Yates, 2005). For the Zöllner illusion, the target of interest includes the main lines and, specifically, their point of convergence. The illusory inducing context consists of the crosshatches—specifically, those that are at a 30° angle. Species that can more readily process the entire visual array would be more likely to perceive this illusory pattern, perhaps contributing to the conflicting results between primates and birds. Again, the different experimental settings between monkey and bird studies might play a certain role too. For instance, viewing distance (Wallace, 1969) is known to partially affect the perception of the illusory pattern in humans. Because monkeys saw the stimuli from larger distances, as compared with birds, it is possible that the different results might be due to this contextual factor. Furthermore, as was mentioned above, these monkeys had previous experience with other illusory phenomena presented in computerized tasks (Agrillo et al., in preparation; Beran & Parrish, 2013). Even though visual illusions previously investigated did not seem to disrupt perception in the objects’ orientation and did not have similar perceptual features, as compared with the stimuli used here, a fairer comparison with birds would need to recruit naïve monkey subjects.
Psychon Bull Rev
An exhaustive debate on this issue extends far beyond the scope of this article. However, regardless of the exact reason underlying these species differences, comparison of our data with the data from previous research in this area indicates the existence of different orientation-tuned mechanisms between the two nonhuman species most frequently adopted in the study of visual illusions, rhesus monkeys and pigeons. In the present study, 30° crosshatches consistently generated a global convergence, leading to an increased perception of converging lines. Future research should attempt to increase potential interfering effects of crosshatches in the judgments of global convergence. Introducing a condition in which crosshatches contradict the global convergence would allow us to assess whether monkeys can show the opposite biasing effect, permitting us to compare the inducing effects of crosshatches when biasing and interfering with the overall convergence of the lines. Also, it is necessary to expand the comparative study of the Zöllner illusion by assessing other mammals (e.g., New World monkeys and nonprimate species) and birds. This will allow us to assess whether the opposite illusory direction reported in primates and birds is a difference between primate and nonprimate species or, more generally, between mammals and birds. Acknowledgments This study was supported by ‘Progetto Giovani Studiosi 2010’ (prot.: GRIC101125) from University of Padova, by FIRB grant (RBFR13KHFS) from ‘Ministero dell'Istruzione, Università e Ricerca’ (MIUR, Italy) to Christian Agrillo, by a 2CI Primate Social Cognition, Evolution and Behavior Fellowship and Duane M. Rumbaugh Fellowship from Georgia State University to Audrey E Parrish, and by grant HD060563 from NICHD to Michael J Beran. The authors thank the three anonymous reviewers for their useful comments and Ted Evans for assistance with data collection.
References Baroncelli, L., Braschi, C., & Maffei, L. (2013). Visual depth perception in normal and deprived rats: Effects of environmental enrichment. Neuroscience, 236, 313–319. Benhar, E., & Samuel, D. (1982). Visual illusions in the baboon (Papio anibis). Animal Learning and Behavior, 10, 115–118. Beran, M. J. (2006). Quantity perception by adult humans (Homo sapiens), chimpanzees (Pan troglodytes), and rhesus macaques (Macaca mulatta) as a function of stimulus organization. International Journal of Comparative Psychology, 19, 386–397. Beran, M. J., & Parrish, A. E. (2013). Visual nesting of stimuli affects rhesus monkeys’ (Macaca mulatta) quantity judgments in a bisection task. Attention Perception and Psychophysics, 75(6), 1243– 1251. Boussaoud, D., Desimone, R., & Ungerleider, L. G. (1991). Visual topography of area TEO in the macaque. Journal of Comparative Neurology, 306, 554–575. Bravo, M., Blake, R., & Morrison, S. (1988). Cats see subjective contours. Vision Research, 28(8), 861–865. Cook, R. G., Qadri, M., Kieres, A., & Commons-Miller, N. (2012). Shape from shading in pigeons. Cognition, 124, 284–303.
Dadda, M., Domenichini, A., Piffer, L., Argenton, F., & Bisazza, A. (2010). Early differences in epithalamic left-right asymmetry influence lateralization and personality of adult zebrafish. Behavioural Brain Research, 20(2), 208–215. Desimone, R., & Gross, C. G. (1979). Visual areas in the temporal cortex of the macaque. Brain Research, 178, 363–380. Evans, T. A., Beran, M. J., Chan, B., Klein, E. D., & Menzel, C. R. (2008). An efficient computerized testing method for the capuchin monkey (Cebus apella): Adaptation of the LRC-CTS to a socially housed nonhuman primate species. Behavior Research Methods, Instruments, & Computers, 40, 590–596. Fujita, K. (1997). Perception of the Ponzo illusion by rhesus monkeys, chimpanzees, and humans: Similarity and difference in the three primate species. Perception and Psychophysics, 59, 284–292. Fujita, K., Blough, D. S., & Blough, P. M. (1991). Pigeons see the Ponzo illusion. Animal Learning and Behavior, 19, 283–293. Huang, X., MacEvoy, S. P., & Paradiso, M. A. (2002). Perception of brightness and brightness illusions in the macaque monkey. Journal of Neuroscience, 22, 9618–9625. Kitaoka, A., & Ishihara, M. (2000). Three elemental illusions determine the Zöllner illusion. Perception and Psychophysics, 62, 569–575. Koffka, K. (1955). Principles of Gestalt Psychology. London: Routledge & Kegan Paul. Nagasaka, Y., Lazareva, O. F., & Wasserman, E. A. (2007). Prior experience affects amodal completion in pigeons. Perception and Psychophysics, 69(4), 596–605. Nakamura, N., Watanabe, S., & Fujita, K. (2008). Pigeons perceive the Ebbinghaus-Titchener circles as an assimilation illusion. Journal of Experimental Psychology: Animal Behavior Processes, 34, 375–387. Nakamura, N., Fujita, K., Ushitani, T., & Miyatat, H. (2006). Perception of the standard and the reversed Muller-Lyer figures in pigeons (Columba livia) and humans (Homo sapiens). Journal of Comparative Psychology, 120, 252–261. Nieder, A. (2002). Seeing more than meets the eye: Processing of illusory contours in animals. Journal of Comparative Physiology A, 188, 249–260. Nieder, A., & Merten, K. (2007). A labeled-line code for small and large numerosities in the monkey prefrontal cortex. Journal of Neuroscience, 27, 5986–5993. Oyama, T. (1975). Determinants of the Zöllner illusion. Psychological Research, 37, 261–280. Parron, C., & Fagot, J. (2007). Comparison of grouping abilities in humans (Homo sapiens) and baboons (Papio papio) with the Ebbinghaus illusion. Journal of Comparative Psychology, 121, 405–411. Parlangeli, O., & Roncato, S. (1995). The global figural characteristics in the Zöllner illusion. Perception, 24, 501–512. Pepperberg, I. M., Vicinay, J., & Cavanagh, P. (2008). Processing of the Müller-Lyer illusion by a grey parrot (Psittacus erithacus). Perception, 37, 765–781. Richardson, W. K., Washburn, D. A., Hopkins, W. D., SavageRumbaugh, S. E., & Rumbaugh, D. M. (1990). The NASA/LRC computerized test system. Behavior Research Methods, Instruments, & Computers, 22, 127–131. Roberts, B., Harris, M. G., & Yates, T. A. (2005). The roles of inducer size and distance in the Ebbinghaus illusion (Titchener circles). Perception, 34, 847–856. Tudusciuc, O., & Nieder, A. (2010). Comparison of length judgments and the Müller–Lyer illusion in monkeys and humans. Experimental Brain Research, 207, 221–231. Wallace, G. (1969). The critical distance of interaction in the Zöllner illusion. Perception and Psychophysics, 5, 261–264. Wallace, G. K., & Moulden, B. (1973). The effect of body tilt on the Zöllner illusion. Quarterly Journal of Experimental Psychology, 25, 10–21. Watanabe, S., Nakamura, N., & Fujita, K. (2011). Pigeons perceive a reversed Zöllner illusion. Cognition, 119, 137–141.
Psychon Bull Rev Watanabe, S., Nakamura, N., & Fujita, K. (2013). Bantams (Gallus gallus domesticus) also perceive a reversed Zöllner illusion. Animal Cognition, 16, 109–115. Wertheimer, M. (1938). Laws of organization in perceptual forms. In W. B. Ellis (Ed.), A sourcebook of Gestalt psychology. San Diego: Harcourt Brace and Company.
White, K. G. (1972). Implicit contours in the Zöllner illusion. American Journal of Psychology, 85, 421–424. Zivotofsky, A. Z., Goldberg, M. E., & Powell, K. D. (2005). Rhesus monkeys behave as if they perceive the Duncker Illusion. Journal of Cognitive Neuroscience, 17, 1011– 1017.
BRIEF REPORT
Do rhesus monkeys (Macaca mulatta) perceive the Zöllner illusion? Christian Agrillo & Audrey E. Parrish & Michael J. Beran
# Psychonomic Society, Inc. 2014
Abstract A long-standing debate surrounds the issue of whether human and nonhuman animals share the same perceptual mechanisms. In humans, the Zöllner illusion occurs when two parallel lines appear to be convergent when oblique crosshatching lines are superimposed. Although one baboon study suggests that they too might perceive this illusion, the results of that study were unclear, whereas two recent studies suggest that birds see this illusion in the opposite direction from humans. It is currently unclear whether these mixed results are an artifact of the experimental design or reflect a peculiarity of birds’ visual system or, instead, a wider phenomenon shared among nonhuman mammals. Here, we trained 6 monkeys to select the narrower of two gaps at the end of two convergent lines. Three different conditions were set up: control (no crosshatches), perpendicular (crosshatches not inducing the illusion), and Zöllner (crosshatches inducing the illusion in humans). During training, the degrees of convergence between the two lines ranged from 15° to 12°. Monkeys that reached the training criterion were tested with more difficult discriminations (11°–1°), including probe trials with parallel lines (0°). The results showed that monkeys perceived the Zöllner illusion in the same direction as humans. Comparison of these data with the data from bird studies points toward the existence of different orientation-tuned mechanisms between primate and nonprimate species.
C. Agrillo (*) Department of General Psychology, University of Padova, Via Venezia 8, 35131 Padova, Italy e-mail: [email protected] A. E. Parrish : M. J. Beran Language Research Center, Georgia State University, Atlanta, USA A. E. Parrish Department of Psychology, Georgia State University, Atlanta, USA
Keywords Visual illusion . Zöllner illusion . Primates . Macaca mulatta . Comparative perception
Introduction The study of visual perception represents one of the research areas that have made the most progress in cognitive sciences during the last 30 years. This is partly due to the increased use of animal models as a tool for understanding the neurocognitive systems that are involved in processing behaviorally relevant information from the retina (e.g., Baroncelli, Braschi, & Maffei, 2013; Cook, Qadri, Kieres, & CommonsMiller, 2012; Dadda, Domenichini, Piffer, Argenton, & Bisazza, 2010). However, despite this growing interest in nonhuman animal visual systems, it is still unclear whether other vertebrates share with humans similar core systems to analyze shape, colors, depth, and the orientation of objects, and it remains unknown to what extent the Gestalt principles (Koffka, 1955; Wertheimer, 1938) described in humans can be generalized to nonhuman animals (Nieder, 2002). Research into visual illusions represents a valuable tool for understanding the mechanisms underlying visual perception, and it can reveal the hidden constraints of the perceptual system. For a long time, the study of visual illusions remained an exclusive area of research for neuroscientists studying humans. But in the last few decades, attention has shifted to include studies with other mammals (e.g., Beran, 2006; Bravo, Blake, & Morrison 1988) and birds (e.g., Nakamura, Watanabe, & Fujita, 2008; Pepperberg, Vicinay, & Cavanagh, 2008). Two animal models, in particular, are frequently adopted in this research field: rhesus monkeys (Macaca mulatta; e.g., Beran & Parrish, 2013; Fujta, 1997; Huang, MacEvoy, & Paradiso, 2002; Zivotofsky, Goldberg, & Powell, 2005) and pigeons (Columba livia; e.g., Cook et al., 2012; Fujita, Blough, & Blough, 1991; Nakamura et al., 2008).
Psychon Bull Rev
One important visual illusion is called the Zöllner illusion. In its classical configuration, two parallel lines appear to be convergent when oblique crosshatching lines are superimposed (see Fig. 1b, column 3). The explanation of this illusion is still uncertain. Several authors believe that the perceptual expansion of acute angles is the main reason underlying this phenomenon (e.g., Kitaoka & Ishihara, 2000; Oyama, 1975; Wallace, 1969). The acute angles formed by the intersections of the lines with the crosshatches would be perceptually enlarged; such expansion would cause a slight perceived change in the orientation of the whole line such that two parallel lines appear to converge. In other words, the illusion would be the result of the sum of the effects of the individual angles (White, 1972). Alternative explanations invoke the role of global characteristics of stimulus configuration (Parlangeli & Roncato, 1995) and viewing distance (e.g., Wallace, 1969). In order to investigate the perception of the Zöllner illusion in nonhuman animals, Watanabe, Nakamura, and Fujita (2011) trained pigeons to peck at the narrower (or wider) of two gaps at the end of a pair of convergent lines. In the test phase, the pigeons were presented with parallel lines with crosshatches, a presentation that typically induces the Zöllner illusion in humans. Watanabe et al. (2011) found that pigeons perceived a form of the illusion, selecting one gap more than would be expected by chance. Interestingly, however, the direction was opposite to that of humans: Pigeons
selected as the narrower gap the one perceived as wider by human subjects. Thus, the stimuli that led to the Zöllner illusion among humans actually improved performance among the pigeons. Watanabe et al. (2011) hypothesized that pigeons underestimated the acute angles between each main line and its crosshatches. The same (reversed) illusion was reported in another bird species as well, the bantam (Gallus gallus domesticus; Watanabe, Nakamura, & Fujita, 2013). Both of these studies raise the intriguing question as to whether the perception of a reversed Zöllner illusion reflects a peculiarity of birds’ visual system or, instead, represents a more pervasive phenomenon of vertebrates’ visual system that somehow is not reflected in the perception of humans. The only attempt to study the Zöllner illusion in mammals was made by Benhar and Samuel (1982). Two baboons were tested in an oddity task in which they were trained to select the odd stimulus in a triad. When presented with two parallel lines (stimulus 1), two convergent lines (stimulus 2), and two “Zöllner” lines (stimulus 3), subjects significantly selected parallel lines as the odd stimulus, presumably showing that the baboons perceived the Zöllner lines as convergent. However, the procedure used in that study did not allow for the establishment of the direction of the illusion in baboons, and so the possibility remained that nonhuman primates might also perceive a reversed Zöllner illusion when an experimental setup similar to those in the recent studies with humans and birds is used.
Fig. 1 Example of testing stimuli. Monkeys were trained to select the narrower gap between the lines’ edges. In the training phase (a), convergences of the main lines ranged from 15° to 12°. In the test phase (b), more difficult discriminations were presented (11°–1°), including parallel
lines. For each phase, three different conditions were presented: control (no crosshatch), perpendicular (90° crosshatches), and Zöllner (30° crosshatches) conditions
Psychon Bull Rev
In the present study, we investigated the perception of the Zöllner illusion in rhesus monkeys. Previous studies showed that rhesus monkeys perceived the Ponzo illusion (Fujta, 1997), the Dunker illusion (Zivotofsky et al., 2005), and the brightness illusion (Huang et al., 2002) in a qualitative fashion that resembles the illusion as seen in human perception, thus making macaques proper candidates for assessing the directionality of any Zöllner illusion in nonhuman animals. In particular, we assessed (1) whether rhesus monkeys perceived this illusory convergence and (2) the direction of the illusion if it occurred. To achieve these goals, monkeys were trained to select the narrower of the two gaps at the end of a pair of convergent lines. Once they reached the learning criterion with easy discriminations, monkeys underwent harder trials, including parallel lines. Three different conditions were set up: control (lines without crosshatches), perpendicular (90° crosshatches to the lines, which result in no illusion for humans), and Zöllner (30° crosshatches to the lines, a presentation that generated the illusory pattern in humans; Wallace & Moulden, 1973; Watanabe et al., 2011). If monkeys perceived the illusory convergence at all, they were expected to perceive parallel lines as converging only in the Zöllner condition, due to the nature of the crosshatches. If that occurred, the direction of their choice would also reveal whether they perceived a humanlike or birdlike (reversed) illusion.
Method Subjects We tested 6 male rhesus macaques from the Language Research Center in Atlanta, GA, including Hank (age 29), Gale (age 29), Luke (age 13), Lou (age 19), Murph (age 19), and Obi (age 9). All monkeys were individually housed with constant visual and auditory access to conspecifics and were given weekly outdoor access with another monkey. Monkeys had constant access to their computerized testing apparatus when they were in the home cage, and thus, they could freely engage tasks to earn food pellets throughout the day. They were fed primate chow, fruits, and vegetables once a day no matter how many experimental trials they completed, and they had 24-h access to water. No animals were food or water deprived for testing purposes. The present study complied with protocols approved by the Georgia State University IACUC and was in full accordance with the USDA Animal Welfare Act and the “Guidelines for the Use of Laboratory Animals.” Materials Testing was conducted via the Language Research Center’s Computerized Test System, consisting of a personal computer,
digital joystick, 17-in. color monitor, and food pellet dispenser (Evans, Beran, Chan, Klein, & Menzel, 2008; Richardson, Washburn, Hopkins, Savage-Rumbaugh, & Rumbaugh, 1990). Monkeys manipulated a digital joystick that generated isomorphic movements of a cursor on the attached screen. The computer program was written using Microsoft Visual Basic 6.0. Food rewards were 94-mg Bio-Serv food pellets that were dispensed contingent upon correct responses. All monkeys had extensive experience with computerized testing, including perceptual illusion experiments (e.g., visually nested stimuli; Beran & Parrish, 2013). Stimuli Stimuli consisted of two black lines that were center justified on the screen. Lines were 89.1 mm long and 0.79 mm thick. Each pattern (the two lines together) was at an angle of approximately 45° with respect to the horizontal axis of the computer screen. Crosshatches intersected the main lines at the crosshatches’ midpoint, were 11.9 mm long and 0.79 mm thick, and were equidistant from each other. We used 10 crosshatches per main line, as was the procedure in the two most recent Zöllner studies (Watanabe et al., 2011, 2013). Crosshatches in the perpendicular condition were placed at a 90° angle to the main lines and a 30° angle to main lines in the Zöllner condition (also consistent with Watanabe et al., 2011, 2013). Therefore, perpendicular and Zöllner conditions presented the same global features (two identical lines and 10 identical crosshatches equally spaced in the two sets). The only difference was the orientation degree of the crosshatches with respect to the main lines. Thus, the Zöllner condition should have exclusively elicited the illusory pattern. However, should the presence of crosshatches have induced any other bias not related to the illusion, this should have occurred in both of these conditions. We presented 31° of convergence (ranging from +15° to −15°), depending upon the phase (see below). Positive and negative degrees varied the side of the smaller gap (right or left) so that each angle was presented equally on both sides. In addition, to prevent the possibility that monkeys could have used other spatial strategies, we randomized the location of the smaller gap so that it appeared in the top and bottom corners of the screen half of the time. We also randomized the side on which the smaller gap appeared, such that it was located on the right and left sides of the screen half of the time. In short, the narrower gap appeared 25 % of the time in the corner “left-above,” 25 % of the time in the corner “leftbelow,” 25 % of the time in the corner “right-above,” and 25 % of the time in the corner “right-below,” thus preventing any possibility of using spatial cues. Normal viewing distance ranged from 30.5 to 40.6 cm, a distance commonly used to investigate visual perception in rhesus monkeys (e.g., Agrillo et al., in preparation; Beran & Parrish, 2013).
Psychon Bull Rev
General procedure
Testing
Each trial presented the monkeys with a cursor that they controlled, the two main lines of variable convergence and crosshatching, and two choice icons (a black leftwardpointing arrow and a black rightward-pointing arrow). The distance between the choice icons was 11 cm. The distance between stimulus and choice icons was 3.5 cm. The arrows were bottom justified on their appropriate side (i.e., the left arrow was left-bottom justified and the right arrow was rightbottom justified). The monkeys controlled a small round cursor that was centered between the choice icons via the joystick. Monkeys moved that cursor into contact with one of the choice icons to indicate which side they perceived to be the side where the main lines converged (i.e., which side had the smaller gap). Selecting the correct choice icon led to one pellet being delivered, whereas selecting the incorrect choice icon led to a 20 s timeout during which the computer screen remained blank. The intertrial interval was 1 s. There was no requirement for a speeded response from the monkeys, since the stimuli would stay on the screen until a monkey made a response. Monkeys were tested in 2 to 4-h sessions at a maximum of once per day. We presented three conditions, including a control condition with no crosshatches, a perpendicular condition with 90° crosshatches, and a Zöllner condition with 30° crosshatches (see Fig. 1). Conditions were randomly presented within a session, with equivalent numbers of each condition per session.
Twenty-three new angles were introduced in the test phase for all three conditions, along with the training stimuli (±15° through ±12), including +11°, +10°, +9°, +8°, +7°, +6°, +5°, +4°, +3°, +2°, +1°, 0°, −11°, −10°, −9°, −8°, −7°, −6°, −5°, −4°, −3°, −2°, and −1°. We included the critical probe trials of the parallel lines (0°) for all three conditions (see Fig. 1b). Again, these probe trials in the Zöllner condition should induce the illusion that parallel lines converge at one end due to the nature of the angled crosshatches. Responses to these probe trials were not reinforced or punished for either response from the monkeys. They simply led directly to the intertrial interval and then the next trial presentation. In order to have a sufficient number of trials to allow us to generate smooth choice curves with the large range of convergence angles presented in the three conditions, the first 8,000 test trials were used in data analysis; monkeys could complete up to 2,000 test trials per session.
Training Eight angles were included in the preliminary training phase for all three conditions, including +15°, +14°, +13°, +12°, −12°, −13°, −14°, and −15°. The criterion for moving to the test phase was 75 % correct choices on each condition in each of two consecutive sessions of at least 250 trials in each condition. If a monkey failed to reach criterion within five sessions, we introduced six larger angles (+30°, +27°, +25°, +22°, +20°, +18°, −18°, −20°, −22°, −25°, −27°, and −30°), a condition called “pretraining 1.” The criterion was the same as above. If a monkey still failed to reach criterion, we introduced only the control condition, in which the bottom main line was parallel with the horizontal axis of the computer screen and, thus, had no pitch (pretraining 2). In that case, the angles were ±15° through ±12°, as above. Criterion was 75 % correct choices in one session of at least 250 trials. Next, they advanced back to the first training phase with angles of ±15° through ±12° for all conditions. If monkeys performed eight consecutive sessions without exhibiting any increase in performance, they were discontinued as subjects in this experiment. Figure 2 summarizes the procedure adopted.
Results Three out of 6 monkeys never reached 75 % correct choices on each condition in two consecutive sessions of the training phase and, hence, did not start the test phase. Hank was discontinued as a subject after pretraining 2, performing a total 7,458 trials, with an average accuracy in the last two sessions of .479. Gale was discontinued as a subject in the pretraining 2 phase, performing a total of 5,041 trials, with an average accuracy in the last two sessions of .442. Obi was discontinued as a subject in the training phase, performing 12,205 trials, with an average accuracy in the last two sessions of .529. Murph, Luke, and Lou reached the learning criterion. During the training phase, Murph performed a total of 5,891 trials, reaching an average accuracy in the last two sessions of .846, .833, and .840 for the control, perpendicular, and Zöllner conditions, respectively. Luke performed a total of 9,175 trials, reaching an accuracy of .875, .830, and .843 for the control, perpendicular and Zöllner conditions, and Lou performed 5,113 trials, with an accuracy of .912, .982, and .908 for the control, perpendicular, and Zöllner conditions. In the test phase, separate two-tailed binomial tests were conducted for the parallel lines presented in each condition. Given the high number of trials per condition, normal approximation to the binomial test (z-scores) was applied. In the Zöllner condition, the performance of all subjects differed from chance levels (Murph N = 156, df = 1, Z = 8.647, p < .05; Lou, N = 196, df = 1, Z = 6.857, p < .05; Luke, N = 107, df = 1, Z = 2.998, p < .05). In all cases, the monkeys’ bias was in the same direction as that reported for humans (Fig. 3).
Psychon Bull Rev
Fig. 2 Schematic representation of the procedure. All monkeys initially started the training phase (1). Subjects reaching the learning criterion began the test phase, which included more difficult discriminations and parallel lines (2a). Those monkeys who did not reach the criterion were, instead, presented with easier discriminations (2b). If they reached the learning criterion, they started again the training phase (1); otherwise,
they were presented with an easier pattern of lines in which the main line was parallel to the horizontal axis (2c). Monkeys were discontinued as subjects if their performance did not show any signs of increasing in proficiency after eight consecutive sessions in the training phase or the pretraining phase
In contrast, no significant choice bias from chance was observed in the control (Murph, N = 193, df = 1, Z = 1.223, p > .05; Lou, N = 216, df = 1, Z = 1.499, p > .05) and perpendicular (Murph, N = 160, df = 1, Z = 0.791, p > .05; Lou, N = 206, df = 1, Z = 0.418, p > .05; Luke, N = 129, df = 1, Z = 1.321, p > .05) conditions, with the exception of Luke’s
performance in the control condition (N = 98, df = 1, Z = 3.302, p < .05). Figure 4 shows the percentage of trials correctly completed as a function of the degree of convergence. For those trials with truly convergent lines, a repeated measures ANOVA with degree of convergence (1–11) and condition (control/perpendicular/Zöllner) as within-subjects factors found main effects of degree of convergence, F(10, 20) = 20.36, p < .001, ηp2 = .816, and condition, F(2, 4) = 8.87, p = .034, ηp2 = .911, indicating that the monkeys’ accuracy improved as the degree of convergence was increased and that their performance varied as a function of condition. Post hoc analyses (LSD) revealed a significant difference between the Zöllner condition and the perpendicular condition (p = .043). No difference was found between the control and perpendicular conditions (p = .422) and between the control and Zöllner conditions (p = .093). A significant interaction also was found, F(20, 40) = 4.74, p < .001, ηp2 = .703, meaning that the slopes of the three conditions were different as a function of the degree of convergence. However, the lack of a significant difference between the control and the Zöllner conditions might be partially due to a potential ceiling effect, considering that easy discriminations were also included in the test phase and that, on these trials,
Fig. 3 Results of test phase, parallel lines: Percentage of trials on which the monkeys selected the end that appears to converge under the illusion as seen by humans. In the Zöllner condition, all monkeys selected the same direction of convergence as that perceived by humans at levels greater than chance
Psychon Bull Rev
Fig. 4 Results of the test phase, convergent lines: Percentage of correct trials plotted against the different degrees of convergence. Overall performance decreases as the degree of convergence decreases. The slopes of the three conditions significantly differ from each other, with an overall
higher performance in the Zöllner condition in the presence of more difficult discriminations (1°–3°). Bars in the average performance represent the standard error of the mean
there was not much need for a facilitation that could occur through the Zöllner illusion. To control for this aspect, we subsequently analyzed whether performance varied among the conditions for only the three smallest degrees of true convergence (1°–3°). Now, there was a main effect of condition, F(2, 4) = 44.42, p < .001, ηp2 = .957. There was no main effect of degree of convergence, F(2, 4) = 6.12, p = .060, ηp2 = .755, and there was no interaction, F(4, 8) = 0.30, p = .871, ηp2 = .130. Post hoc analyses (LSD) showed that better performance occurred in the Zöllner condition, as compared with the two other conditions: Zöllner versus control, p = .010; Zöllner versus perpendicular, p = .025; control versus perpendicular, p = .890.
with parallel lines. The present results indicated that monkeys perceived the Zöllner illusion in the same direction as humans. When there was no correct answer because the lines were parallel and there were 30° crosshatches as are typically used in presenting the Zöllner illusion, all monkeys were biased to indicate that the lines converged toward the end for which humans also perceive parallel lines as being more convergent. When we analyzed the performance in the presence of convergent lines, accuracy decreased with decreases in the degree of convergence. However, a different pattern was found as a function of stimulus type. In the Zöllner condition, the percentage of correct choices was less negatively affected by the degree of convergence. In particular, the monkeys’ overall accuracy was higher than that reported in the perpendicular condition, a condition characterized by the same overall stimulus features. Since the degree of convergence of crosshatches is the only difference between the two conditions, the different performance in the Zöllner condition can be ascribed only to this variable: 30° crosshatches led to an increased perception of convergence between the lines, thus increasing the monkeys’ ability to select the narrower gap. Not only do these results together indicate that rhesus monkeys perceive the
Discussion The present study assessed whether rhesus monkeys perceived the Zöllner illusion. Monkeys were trained with convergent lines having no crosshatches (control condition), 90° crosshatches (perpendicular condition), and 30° crosshatches (Zöllner condition) and then were observed in probe trials
Psychon Bull Rev
Zöllner illusion, but also this study confirms that they too show the illusion in the same direction as humans, rather than the reverse pattern that normally is reported in birds (Watanabe et al., 2011, 2013). It is interesting to note that 3 subjects out of 6 did not reach the learning criterion. Unfortunately, we can only speculate about the reasons underlying their lack of discrimination, since they otherwise have performed comparably on other perceptual and cognitive tasks to the macaques that were successful in this study. One possibility is that these subjects were not particularly skillful in visual discrimination or were not motivated to accomplish the task. Two of these monkeys were the oldest individuals of the sample, a factor that might be partially related to lack of motivation or poorer visual acuity. Also, we must acknowledge the possibility that previous computerized tasks might have partially affected the performance: All monkeys had previous experience with numerous cognitive-perceptual computerized tasks. Those past experiences might have generated interference in their ability to learn the basic parameters of this task. One might also argue that the lack of learning to perform this discrimination might be due to specific difficulties in discriminating visual stimuli with the particular characteristics that are relevant to assessing this visual line illusion, such as converging lines with crosshatches. However, we do not feel that this is the case, since similar (low) performances were found for the control, perpendicular, and Zöllner conditions in the last sessions of their training phase. Previous studies have shown that humans and rhesus monkeys share similar Gestalt principles in their perceptual experiences (Fujta, 1997; Tudusciuc & Nieder, 2010; Zivotofsky et al., 2005). Our study aligns with this literature, since the Zöllner illusion emerges due to the global processing and subsequent interaction of all elements within the visual array, including main lines and crosshatches. These and similar findings indicate the value of using rhesus monkeys as a model to investigate the mechanisms underlying objects’ orientation too. For instance, the cortical areas involved in the expansion of acute angles in Zöllner patterns are partly unknown. It has been hypothesized (Kitaoka & Ishihara, 2000) that there is an involvement of areas V1, V2, and V4, as well as higher areas such as the inferotemporal cortex (TE). Rhesus monkeys are commonly used to study neuroanatomical correlates of perceptual and cognitive functions in primates (Desimone & Gross, 1979; Nieder & Merten, 2007), and their visuotopic organization of visual areas from V1 through V4 and of area TE have been extensively mapped (Boussaoud, Desimone, & Ungerleider, 1991). In this sense, future macaque brain imaging studies could help us to shed light into the neural networks involved in the Zöllner illusion. This study also raises the intriguing question about why humans and rhesus monkeys perceive the Zöllner illusion differently from pigeons and bantams. With respect of this
topic, Watanabe et al. (2011) referred to the contrast and assimilation hypothesis advanced by Nakamura et al. (2008). According to this hypothesis, birds would experience both “assimilation” (the phenomenon according to which the perception of two figures tends to assimilate and one figure looks similar to the other in its perceptual features) and “contrast” (the opposite phenomenon according to which the perceptual differences between two figures are enhanced) effects. Birds, however, would be less sensitive to contrast effects, as compared with humans (Watanabe et al., 2011) and, presumably, rhesus monkeys. As a consequence, they would underestimate the acute angles between lines and crosshatches in the Zöllner pattern as a result of assimilation of orientation of two lines, leading to a reversed Zöllner illusion. In line with these findings, previous studies showed that pigeons display different perceptual mechanisms, as compared with those described in primates: Pigeons do not perceive some visual illusions in the same way as humans (e.g., Ebbinghaus-Titchener and Müller-Lyer illusions; Nakamura, Fujita, Ushitani, & Miyatat, 2006; Nakamura et al., 2008), and there is also debate as to whether they can exhibit amodal completion (Fujita 2004; Nagasaka, Lazareva, & Wasserman, 2007). The Zöllner illusion in our species is affected by global characteristics of stimulus configuration (e.g., Parlangeli & Roncato, 1995), and the possibility remains that monkeys might be more attentive to global properties of the images than are pigeons/chickens. Geometric illusions, like the Zöllner illusion, are critically dependent upon the simultaneous processing of the entire visual array, including the illusory inducing context and the target stimulus (e.g., Parron & Fagot, 2007; Roberts, Harris, & Yates, 2005). For the Zöllner illusion, the target of interest includes the main lines and, specifically, their point of convergence. The illusory inducing context consists of the crosshatches—specifically, those that are at a 30° angle. Species that can more readily process the entire visual array would be more likely to perceive this illusory pattern, perhaps contributing to the conflicting results between primates and birds. Again, the different experimental settings between monkey and bird studies might play a certain role too. For instance, viewing distance (Wallace, 1969) is known to partially affect the perception of the illusory pattern in humans. Because monkeys saw the stimuli from larger distances, as compared with birds, it is possible that the different results might be due to this contextual factor. Furthermore, as was mentioned above, these monkeys had previous experience with other illusory phenomena presented in computerized tasks (Agrillo et al., in preparation; Beran & Parrish, 2013). Even though visual illusions previously investigated did not seem to disrupt perception in the objects’ orientation and did not have similar perceptual features, as compared with the stimuli used here, a fairer comparison with birds would need to recruit naïve monkey subjects.
Psychon Bull Rev
An exhaustive debate on this issue extends far beyond the scope of this article. However, regardless of the exact reason underlying these species differences, comparison of our data with the data from previous research in this area indicates the existence of different orientation-tuned mechanisms between the two nonhuman species most frequently adopted in the study of visual illusions, rhesus monkeys and pigeons. In the present study, 30° crosshatches consistently generated a global convergence, leading to an increased perception of converging lines. Future research should attempt to increase potential interfering effects of crosshatches in the judgments of global convergence. Introducing a condition in which crosshatches contradict the global convergence would allow us to assess whether monkeys can show the opposite biasing effect, permitting us to compare the inducing effects of crosshatches when biasing and interfering with the overall convergence of the lines. Also, it is necessary to expand the comparative study of the Zöllner illusion by assessing other mammals (e.g., New World monkeys and nonprimate species) and birds. This will allow us to assess whether the opposite illusory direction reported in primates and birds is a difference between primate and nonprimate species or, more generally, between mammals and birds. Acknowledgments This study was supported by ‘Progetto Giovani Studiosi 2010’ (prot.: GRIC101125) from University of Padova, by FIRB grant (RBFR13KHFS) from ‘Ministero dell'Istruzione, Università e Ricerca’ (MIUR, Italy) to Christian Agrillo, by a 2CI Primate Social Cognition, Evolution and Behavior Fellowship and Duane M. Rumbaugh Fellowship from Georgia State University to Audrey E Parrish, and by grant HD060563 from NICHD to Michael J Beran. The authors thank the three anonymous reviewers for their useful comments and Ted Evans for assistance with data collection.
References Baroncelli, L., Braschi, C., & Maffei, L. (2013). Visual depth perception in normal and deprived rats: Effects of environmental enrichment. Neuroscience, 236, 313–319. Benhar, E., & Samuel, D. (1982). Visual illusions in the baboon (Papio anibis). Animal Learning and Behavior, 10, 115–118. Beran, M. J. (2006). Quantity perception by adult humans (Homo sapiens), chimpanzees (Pan troglodytes), and rhesus macaques (Macaca mulatta) as a function of stimulus organization. International Journal of Comparative Psychology, 19, 386–397. Beran, M. J., & Parrish, A. E. (2013). Visual nesting of stimuli affects rhesus monkeys’ (Macaca mulatta) quantity judgments in a bisection task. Attention Perception and Psychophysics, 75(6), 1243– 1251. Boussaoud, D., Desimone, R., & Ungerleider, L. G. (1991). Visual topography of area TEO in the macaque. Journal of Comparative Neurology, 306, 554–575. Bravo, M., Blake, R., & Morrison, S. (1988). Cats see subjective contours. Vision Research, 28(8), 861–865. Cook, R. G., Qadri, M., Kieres, A., & Commons-Miller, N. (2012). Shape from shading in pigeons. Cognition, 124, 284–303.
Dadda, M., Domenichini, A., Piffer, L., Argenton, F., & Bisazza, A. (2010). Early differences in epithalamic left-right asymmetry influence lateralization and personality of adult zebrafish. Behavioural Brain Research, 20(2), 208–215. Desimone, R., & Gross, C. G. (1979). Visual areas in the temporal cortex of the macaque. Brain Research, 178, 363–380. Evans, T. A., Beran, M. J., Chan, B., Klein, E. D., & Menzel, C. R. (2008). An efficient computerized testing method for the capuchin monkey (Cebus apella): Adaptation of the LRC-CTS to a socially housed nonhuman primate species. Behavior Research Methods, Instruments, & Computers, 40, 590–596. Fujita, K. (1997). Perception of the Ponzo illusion by rhesus monkeys, chimpanzees, and humans: Similarity and difference in the three primate species. Perception and Psychophysics, 59, 284–292. Fujita, K., Blough, D. S., & Blough, P. M. (1991). Pigeons see the Ponzo illusion. Animal Learning and Behavior, 19, 283–293. Huang, X., MacEvoy, S. P., & Paradiso, M. A. (2002). Perception of brightness and brightness illusions in the macaque monkey. Journal of Neuroscience, 22, 9618–9625. Kitaoka, A., & Ishihara, M. (2000). Three elemental illusions determine the Zöllner illusion. Perception and Psychophysics, 62, 569–575. Koffka, K. (1955). Principles of Gestalt Psychology. London: Routledge & Kegan Paul. Nagasaka, Y., Lazareva, O. F., & Wasserman, E. A. (2007). Prior experience affects amodal completion in pigeons. Perception and Psychophysics, 69(4), 596–605. Nakamura, N., Watanabe, S., & Fujita, K. (2008). Pigeons perceive the Ebbinghaus-Titchener circles as an assimilation illusion. Journal of Experimental Psychology: Animal Behavior Processes, 34, 375–387. Nakamura, N., Fujita, K., Ushitani, T., & Miyatat, H. (2006). Perception of the standard and the reversed Muller-Lyer figures in pigeons (Columba livia) and humans (Homo sapiens). Journal of Comparative Psychology, 120, 252–261. Nieder, A. (2002). Seeing more than meets the eye: Processing of illusory contours in animals. Journal of Comparative Physiology A, 188, 249–260. Nieder, A., & Merten, K. (2007). A labeled-line code for small and large numerosities in the monkey prefrontal cortex. Journal of Neuroscience, 27, 5986–5993. Oyama, T. (1975). Determinants of the Zöllner illusion. Psychological Research, 37, 261–280. Parron, C., & Fagot, J. (2007). Comparison of grouping abilities in humans (Homo sapiens) and baboons (Papio papio) with the Ebbinghaus illusion. Journal of Comparative Psychology, 121, 405–411. Parlangeli, O., & Roncato, S. (1995). The global figural characteristics in the Zöllner illusion. Perception, 24, 501–512. Pepperberg, I. M., Vicinay, J., & Cavanagh, P. (2008). Processing of the Müller-Lyer illusion by a grey parrot (Psittacus erithacus). Perception, 37, 765–781. Richardson, W. K., Washburn, D. A., Hopkins, W. D., SavageRumbaugh, S. E., & Rumbaugh, D. M. (1990). The NASA/LRC computerized test system. Behavior Research Methods, Instruments, & Computers, 22, 127–131. Roberts, B., Harris, M. G., & Yates, T. A. (2005). The roles of inducer size and distance in the Ebbinghaus illusion (Titchener circles). Perception, 34, 847–856. Tudusciuc, O., & Nieder, A. (2010). Comparison of length judgments and the Müller–Lyer illusion in monkeys and humans. Experimental Brain Research, 207, 221–231. Wallace, G. (1969). The critical distance of interaction in the Zöllner illusion. Perception and Psychophysics, 5, 261–264. Wallace, G. K., & Moulden, B. (1973). The effect of body tilt on the Zöllner illusion. Quarterly Journal of Experimental Psychology, 25, 10–21. Watanabe, S., Nakamura, N., & Fujita, K. (2011). Pigeons perceive a reversed Zöllner illusion. Cognition, 119, 137–141.
Psychon Bull Rev Watanabe, S., Nakamura, N., & Fujita, K. (2013). Bantams (Gallus gallus domesticus) also perceive a reversed Zöllner illusion. Animal Cognition, 16, 109–115. Wertheimer, M. (1938). Laws of organization in perceptual forms. In W. B. Ellis (Ed.), A sourcebook of Gestalt psychology. San Diego: Harcourt Brace and Company.
White, K. G. (1972). Implicit contours in the Zöllner illusion. American Journal of Psychology, 85, 421–424. Zivotofsky, A. Z., Goldberg, M. E., & Powell, K. D. (2005). Rhesus monkeys behave as if they perceive the Duncker Illusion. Journal of Cognitive Neuroscience, 17, 1011– 1017.
