Do Animals Have Insight, and What Is Insight Anyway?
Shettleworth, Sara J., Canadian Journal of Experimental Psychology
We cannot test animals for insight's distinctive phenomenology, the "aha" experience, but we can study the processes underlying insightful behaviour, classically described by Köhler as sudden solution of a problem after an impasse. The central question in the study of insightful behaviour in any species is whether it is the product of a distinctive cognitive process, insight. Although some claims for insight in animals confuse it with other problem-solving processes, contemporary research on string pulling and other physical problems, primarily with birds, has uncovered new examples of insightful behaviour and shed light on the role of experience in producing it. New research suggests insightful behaviour can be captured in common laboratory tasks while brain activity is monitored, opening the way to better integration of research on animals with the cognitive neuroscience of human insight.
Keywords: insight, comparative cognition, problem solving, animals, learning
A chimpanzee gestures fruitlessly toward a banana lying beyond arm's reach. An elephant extends its trunk toward an apple hanging high overhead. When animals are confronted with inaccessible rewards like these, do they ever use insight to solve the problem of getting them? That is, do they seem to suddenly see a solution? This question was central to some of the earliest studies of animal problem solving (Boakes, 1984). It also motivated the seminal work of Köhler (1925/1959) and figured in Thorpe's (1956) discussion of learning in animals, but otherwise the question whether nonhuman species have insight was largely neglected by comparative psychologists for most of the 20th century. One exception was D. O. Hebb (1949), in whose honour this essay is written and whose contributions are discussed later in this article.
Recently the study of animal insight has reemerged as part of the burgeoning field of comparative cognition. Contemporary research on many topics in this field such as memory, numerical and spatial cognition, and theory of mind has become seamlessly integrated with theory and data from cognitive neuroscience and human cognitive and developmental psychology (see Shettleworth, 2010a). A good example is in the study of numerical cognition. Parallel experiments with monkeys and children have uncovered common systems for precise assessment of small numbers of items and for approximate representations of large quantities (Cantlon, Piatt, & Brannon, 2009). These same systems are evident in behavioural tests with human adults and in single-cell recordings with monkeys (Nieder & Dehaene, 2009), supporting a view of human numerical competence in which it develops via uniquely human mechanisms from simple core systems shared with other species (Spelke & Lee, 2012). In contrast, with few exceptions (e.g., Call, in press), most current discussions of animal insight almost completely overlook recent literature on human problem-solving. They also tend to give short shrift to explanations of insightful behaviours in terms of basic learning mechanisms. The purpose of this article is to redress this balance and to ask whether doing so suggests new directions for comparative research.
The article has three main sections. For a basis from which to compare work with nonhuman animals (henceforth simply animals), we first look at major issues and approaches in current research on human insight. We then turn to the landmarks in 20th century research on animal insight, with particular attention to the work of Köhler (1925/1959) and Epstein (1985). Finally we look at a sample of 21st century studies with diverse species and paradigms that have been claimed to reveal insight.
Human Insight: A Brief Review
Like the problem of the out-of-reach banana or the dangling apple, problems that humans may solve insightfully involve a goal that cannot be reached in the way most subjects try first. Such problems can be as simple as a paper-and-pencil test in which dots in a 3 X 3 array must be joined by four connected straight lines (the "9 dot" problem, see Weisberg, 2006). …