The Parrot Brain On Shapes: Similarities with Human Visual Processing

Introduction

Objects are often not fully visible in everyday life. Human beings are capable of processing the complex visual information related to “incompleteness” because our visual environment is primarily composed of opaque objects that can overlap and partially hide each other (e.g., Pepperberg and Nakayama 2016). Scientists believe that many nonhuman species are also able to deal with “incompleteness”. For instance, processing partial clues about a potential predator and reacting is safer than not, even when false alarms arise (Fig 1).

Objects are not always fully visible in everyday life, yet the ability to process partial clues, like a potential predator, can be crucial in everyday life.

Figure 1. Objects are not always fully visible in everyday life, yet the ability to process partial clues, like a potential predator, can be crucial in everyday life. Photo credit: Tony Alter via Flickr Creative Common

Amodal completion

When a partially covered object is still easily seen and identified, like a cat in grass or a square behind a circle (Fig 2), this process is an example of amodal completion (e.g., Pepperberg and Nakayama 2016). Whereas previous knowledge and memory generally play a role in the recognition of non-occluded objects, additional perceptual processes (e.g., an understanding of depth perception, that the circle is in front of the square) seem to be required for amodal completion (Vallortigara 2006).

square behind circle

Figure 2. When a partially covered object is still easily seen and identified, this is described as amodal completion. Despite the missing lower right corner, the partially occluded object is quickly recognizable as a square.

A number of species have been shown to perceive something about occluded objects, including chicks, mynahs, magpies and monkeys (Vallortigara 2006; reviewed in Pepperberg and Nakayama 2016, Plowright et al 1998, Funk 1996, Pepperberg and Funk 1990). Circumstantial evidence for amodal completion has also been recently evaluated by testing the reaction in wild birds (Paridae spp.) to occluded or amputated models of a predator (Accipiter nisus) attached to a feeder (Tvardíková and Fuchs 2010). However, alternative explanations exist for the behavior of all these species: the design of their tasks were such that subjects might be focusing on simple aspects of the stimuli without understanding what they were really seeing, for example, matching the 90-degree angle in the upper left of Figure 1 to one in a test shape without necessarily recognizing that the figure is actually an occluded square or, in the case of the wild birds, having experience with a partially hidden raptor that flies out for an attack.

Nevertheless, the concept of amodal completion appears to be an ecologically valid and ubiquitous task. For instance, recognition of a partially occluded mother would be useful when you can move by yourself to rejoin her in order to reinstate social contact. This is probably why recognition of partly occluded objects emerges early in precocial chicks but not for highly altricial species like the human newborn (Vallortigara 2006).

Pigeons, and some other animals, may have evolved with a visual system that does not actively complete occluded stimuli, but recognizes them as two separate figures (Fig 3) (Fujita 2006). Pigeons appear to respond to visual stimuli on the basis of local, visible features, and fail to complete—or possibly even perceive—continuation of the figure behind the occluded object (Fujita 2006). Moreover, completion theoretically requires more processing time. It may be adaptive for pigeons not to complete stimuli, at least not in the feeding tasks commonly used in their tests (Fujita 2006).

Pigeons have been shown not to complete partially occluded figures in studies using a variety of stimuli and procedures

Figure 3. Pigeons have been shown not to complete partially occluded figures in studies using a variety of stimuli and procedures and view them as two items that, when separated, appear as in this figure.

Plowright et al (1998) found that pigeons lost interest in food when it became invisible behind a screen. Fujita 2001 hypothesized that this was related to this species’ nutritional strategy. Pigeons are grain eaters, and because grain is usually abundant, the animal is not required to search behind obstacles. Amodal completion may be important for other avian species that engage in finding and eating a wider variety of food items, including worms and insects that often hide under leaves or soil and may be only partly visible.

 

Modal completion

A different type of visual task involves subjective contours, or seeing figures that are actually imaginary. The formal term is “modal completion”. The most well-known example of modal completion is the Kanizsa triangle, in which one can see a triangle bounded by three “pac-men” figures positioned in such a way that the open angles of 60-degrees all point inwards to the same region (Fig 4). In addition to the subjective or illusory contours that create the edges of a triangle, a second illusory component consists of the triangle appearing brighter than the surrounding region even though it has the same physical luminance. Bees and barn owls, among other species, have also been shown to perceive something about subjective contours, but for the same reasons described for amodal completion tasks, the data are ambiguous (reviewed in Pepperberg and Nakayama, 2016).

Subjective or illusory contours are visual illusions that evoke the perception of an edge without a difference in luminance, color, or texture across that edge.

Figure 4. Subjective or illusory contours are visual illusions that evoke the perception of an edge without a difference in luminance, color, or texture across that edge.

The experiment

We thus decided to see what would happen if we could test a nonhuman exactly the same way that humans are tested: by showing stimuli as in the figures above, and simply asking what is seen. We had the perfect subject: Griffin, a 16-year old, male Grey parrot (Psittacus erithacus), who was previously taught English-language labels for colors and shapes of various solid objects (Fig 5). Unlike many animal subjects, Griffin could vocally describe the items in his environment, offering a unique opportunity to compare human and nonhuman data (Pepperberg and Nakayama 2016). He had, however, never been tested on two-dimensional pictures of objects.

Griffin the grey parrot used his abilities to produce labels for colors and shape

Figure 5. Griffin the Grey parrot (Psittacus erithacus) used his abilities to produce labels for colors and shapes (calling them 1, 2, 3, 4, 6, 8-corner). Eight-cornered shapes were excluded in testing.

The experiment was designed to answer three questions (Pepperberg and Nakayama 2016). Can a Grey parrot…

  • Recognize two-dimensional stimuli after having been trained on three-dimensional objects?
  • Perform identification tasks using novel images to show the presence of modal and amodal completion?
  • Perform recognition tasks on objects for which he has not been trained and under very different circumstances from his training?

Despite the lack of training on two-dimensional items such as those in Figures 4, 6 and 7, as well as having limited exposure to occluded objects and none to subjective contours, Griffin immediately transferred from training on three-dimensional objects to testing on two-dimensional stimuli. Griffin could perceive and appropriately label occluded and Kanizsa figures, thereby demonstrating both amodal and modal completion (Pepperberg and Nakayama 2016).

Each of the 2D stimuli were held vertically, approximately 15-20 cm from one of Griffin’s eyes to attract his attention (Video 1). Monocular vision was used because the extent of binocular overlap in the Grey parrot is unknown. The Senegal parrot, a species that is not very closely related to Greys, has approximately 30% binocular overlap (Demery et al 2011).

Video 1. Two-dimensional stimuli were held vertically towards one of Griffin’s eyes. The experimenter manually tracked the position of the test object with respect to the parrot’s head to maintain presentation in front of one eye.

 

To test for amodal completion, Pepperberg’s students asked Griffin to label various colored polygons occluded by black circles (Fig 6). For evaluation of modal completion, Griffin was shown Kanizsa figures constructed using black “pac-men” to form regular polygons on colored paper (Fig 7) (Pepperberg and Nakayama 2016).

“What shape blue?” was one question used to test Griffin for amodal completion

Figure 6. “What shape blue?” was one question used to test Griffin for amodal completion. Note: No Arabic numbers were on Griffin’s stimuli. We used 38 different stimuli, including five probes (one for each possible shape) of the type shown in Figure 3.

 

Nine different Kanizsa figures were used in this study

Figure 7. Twenty-eight different Kanizsa figures were used in this study, plus nine probes of the type shown in the right of this figure. Griffin was asked to describe how many corners were present in the object created by “pac-men” figures. Note: Most experiments in the non-human literature use one identical figure for testing. Again, no numbers were present on Griffin’s stimuli.

The colors used were those tested in a previous study to ensure Griffin would interpret the label for these printed colors correctly (Pepperberg et al 2008). Orange and purple were adjusted within appropriate ranges as these colors can be problematic for Greys to identify (Pepperberg 1994; 2006). The ability of psittacine birds to see in the ultraviolet spectrum can cause their color perception to differ somewhat from that of humans (Pepperberg and Nakayama 2016, Carvalho et al 2011, Goldsmith and Butler 2005, Bowmaker et al 1994).

Test results showed high accuracy (Table 1). Griffin was correct on the first trial of each modal and amodal task, which suggested he immediately transferred his responses—without training—from non-occluded three-dimensional items to drawings of occluded or imaginary shapes. Griffin responded to the true shape of figures, despite the fact that all were partially hidden or not physically real. Most errors occurred for one-and three-cornered polygons and triangles, which he seemed to confuse in both modal and amodal tasks (Pepperberg and Nakayama 2016). However, these errors were few overall, and did not suggest that he was focusing solely on the 60-degree angles involved.

Table 1. Summary of Griffin’s test results
Parameter measuredPercent accuracy *# of trials correct
Modal completion7629/38
Amodal completion7023/33
*Chance is associated with 20% accuracy

Conclusion

In the human visual experience, when an object is partially concealed by an obstacle, we do not perceive only the pieces or fragments of that object. The parts that are directly visible usually suffice for recognition of the whole object. Humans also tend to see objects that are created by illusions, as in the Kanizsa figures. At least one Grey parrot, Griffin, responds similarly and these results—particularly his transfer from solid three-dimensional figures to occluded and imaginary two-dimensional stimuli, without any training—imply that he has a sophisticated concept of shape. Although this information itself is unlikely to be directly applicable when planning environmental enrichment, it reminds us of the intelligence of these birds and the need for such stimulation in their environment.

References

References

Bowmaker JK, Heath LA, Das D, Hunt DM. Spectral sensitivity and opsin structure of avian rod and cone visual pigments. Invest Ophthalmol Vis Sci 35:1708, 1994.

Carvalho LS, Knott B, Berg ML, et al. Ultraviolet-sensitive vision in long-lived birds. Proc Royal Society B 278:107-114, 2011. Available at http://rspb.royalsocietypublishing.org/content/278/1702/107. Accessed July 28, 2016.

Fujita K. Seeing what is not there:  Illusion, completion, and spatiotemporal boundary formation in comparative perspective. In: Wasserman EA, Zentall TR (eds). Comparative Cognition:  Experimental Explorations of Animal Intelligence. New York, NY: Oxford University Press; 2006: 37, 42-50.

Fujita K. Perceptual completion in rhesus monkeys (Macaca mulatta) and pigeons (Columba livia). Perception and Psychophysics 63:115-125, 2001.

Funk MS. Development of object permanence in the New Zealand parakeet (Cyanoramphus auriceps). Animal Learning and Behavior 24:375-383, 1996.

Goldsmith T, Butler BK. Color vision of the budgerigar (Melopsittacus undulates): Hue matches, tetrachromacy, and intensity discrimination. J Comp Physiol 191:933-951, 2005. Available at http://www.2ndchance.info/tameparrot-Goldsmith2005.pdf. Accessed July 28, 2016.

Pepperberg IM. Addition by a Grey parrot (Psittacus erithacus), including absence of quantity. J Comp Psychol 120:1-11, 2006.

Pepperberg IM. The Alex Studies; Cognitive and Communicative Abilities of Grey Parrots. Cambridge, MA: Harvard; 1999.

Pepperberg IM. Evidence for numerical competence in an African Grey parrot (Psittacus erithacus). J Comp Psychol 106:295-305, 1994.

Pepperberg IM, Funk MS. Object permanence in four species of psittacine birds:  An African Grey parrot (Psittacus erithacus), an Illiger mini maaw (Ara maracana), a parakeet (Melopsittacus undulatus), and a cockatiel (Nymphicus hollandicus). Animal Learning and Behavior 18:97-108, 1990

Pepperberg IM, Nakayama K. Robust representation of shape in a Grey parrot (Psittacus erithacus). Cognition 153:146-160, 2016.

Pepperberg IM, Vicinay J, Cavanagh P. The Muller-Lyer illusion is processed by a Grey parrot (Psittacus erithacus). Perception 37:765-781, 2008.

Plowright CMS, Reid S, Kilian T. Finding hidden food: Behavior on visible displacement tasks by mynahs (Gracula religiosa) and pigeons (Columba livia). J Comp Psychology 112:13-25, 1998.

Tvardíková K, Fuchs R. Tits use amodal completion in predator recognition: A field experiment. Animal Cognition 13:609-615, 2010.

Vallortigara G. The cognitive chicken:  Visual and spatial cognition in a nonmammalian brain. In: Wasserman EA, Zentall TR (eds). Comparative Cognition:  Experimental Explorations of Animal Intelligence. New York, NY: Oxford University Press; 2006: 53-70.

To cite this page:

Pepperberg I, Pollock C. The parrot brain on shapes: Similarities with human visual processing. LafeberVet web site. Available at  https://lafeber.com/vet/parrot-brain-shapes-similarities-human-visual-processing/