Concepts, Comparative Perspectives

Do Animals Have Concepts?

It is a longstanding empirical question in the comparative literature whether animals similarly form concepts. Richard Herrnstein and colleagues pioneered some of the earliest studies with pigeons when they asked whether the birds could learn to discriminate between pictures of humans and pictures that did not include humans—thus indicating that they had a concept of human. The researchers then went on to show that pigeons also had concepts for water and trees and other natural objects. The early concept formation literature was thus dominated by studies aiming to show that animals formed many natural concepts, at many levels of abstraction. Following in the important footsteps of Eleanor Rosch's seminal work with human children, it was important to show that animals could form abstract as well as concrete level categories. Very concrete categories—those in which the exemplars all share many physical features in common—can be discriminated solely on the basis of attending to perceptual features. If animals respond differently to pictures of humans versus nonhumans solely by looking for and responding to skin-colored patches or eyes or clothing, this would not indicate that they truly held a concept of human, but rather that they responded to the world as a conglomeration of perceptual features, not as conceptual wholes. However, if they could form categories for things that were not bound together by perceptual features, such as functional categories such as foods and tools and toys and animals—categories that can include exemplars that are perceptually quite diverse—or even more abstract categories that include nonobservable entities such as “things that make one happy”—this would constitute evidence that animals also form concepts.

Methods of Study

Thus, comparative researchers interested in the general topic of whether animals form concepts have focused on diverse but convergent questions. Some have continued in the tradition of the early research, presenting animals with various pictorial stimuli and asking them to categorize the stimuli, using two-choice discrimination paradigms, in which they select images that belong to one category, and not images that belong to another, or simply that don't belong. Alternatively, they might use a match-to-sample paradigm, in which animals select from comparison images the one that matches the category depicted in a sample image. Experimenters can then rigorously manipulate and examine the stimuli chosen and rejected by their subjects in order to determine what features are being used to drive their choices. Recently experimenters have devised new techniques, such as bubbles, for isolating and illuminating salient features of the stimuli to present and measuring the impact on the animals' choices. Alternatively, researchers can now use more flexible software to directly manipulate the features in a stimulus to examine the individual effects of, for example, enlarged eyes, elongated legs, darkened hair, and so on. While it is difficult to adequately summarize the outcome of the various studies across all of the species studied, suffice to say that most animals use a variety of salient perceptual features to aid them in performing these categorical discrimination tasks—some of which are categorically relevant, such as the presence of eyes when discriminating animals from nonanimals, and some not categorically relevant, such as preferring close-up to distant images when discriminating humans from nonhumans. Thus, this large body of literature remains agnostic on the question of whether animals truly form concepts in the same manner that humans do, independent of specific individual features but is suggestive that animals are much more perceptually biased rather than forming overarching abstract concepts.

...