Abstract Concept Formation
Concepts are formed when an organism is capable of representing a mental construct that subsumes multiple related exemplars. Exemplars may be related by virtue of shared physical features, functions, traits, capacities, relationships, or other unifying characteristics. Abstract concepts are representations of conceptual categories containing exemplars that are not strictly tied to observable features of the stimuli and that typically subsume several more concrete categories. Forming such concepts requires inference or generalization from observed features to the construct that ties exemplars together.
The idea of abstraction is somewhat nebulous, and evidence for such in nonhuman minds is somewhat elusive. It has been suggested that most animals are not capable of abstraction by both historic (Locke 1690/1975) and more recent figures (Mackintosh 2000), and even those that believe some animals capable of some degree of abstraction have placed limits on their capacities, stipulating that they may not be capable of reasoning about certain types of relations (Penn et al. 2008) or unobservables (Vonk and Povinelli 2006). Part of the difficulty in providing evidence for abstraction in nonhumans stems from the challenge in defining what it means to be abstract in the first place; conventional definitions focus on the idea that abstract concepts require generalization beyond strictly observable features to include constructs for unobservables such as traits, functions, and causal forces.
As such, an abstract concept can encompass qualities that are associated by virtue of underlying causal forces, thematic representations, or relations between items. Vonk and Povinelli (2006) focused on the notion of unobservables to represent a class of objects that are strictly hypothetical and cannot take on physical form: constructs such as thoughts, memories, ideas, beliefs and physical forces, such as gravity. The numerical concept of zero, or the absence of something, can also be considered abstract, as can the idea of sameness or difference, which pertains to the relationship between objects. Abstraction is required when the relationship between exemplars within an overarching category must be inferred based on observable qualities rather than derived directly from them. That is, having a concept of sameness goes beyond matching something orange to something else orange based on perceived perceptual similarity to abstracting the general relation “same” that could potentially be generalized to two green items or two items of the same shape or material.
Experimental Methods to Study Abstraction
The human capacity for language allows for representational content to be communicated. Humans have words for constructs that cannot be visualized, such as thoughts, physical forces, and language itself. Nonhumans do not have the luxury of communicating their thoughts in symbolic form, and therefore the content of their minds must be inferred through categorization tasks, such as matching-to-sample (MTS) and two alternative forced choice tests (2AFC), where they are required to match a comparison stimulus to a sample based on a perceived relation (similarity as in traditional identity MTS or conceptual MTS tasks, or difference, as in oddity MTS tasks), or to choose members among pairs of items that belong to the same conceptualized category (2AFC tasks). Nonhumans are quite skilled at perceiving perceptual similarity and categorizing items that share qualities such as shape or color (Vonk 2003), but there is less evidence for their ability to categorize on the basis of more abstract qualities, such as social relationships (Vonk 2002; Vonk and Johnson–Ulrich 2014). One area in which many nonhumans excel is numerosity, which has been suggested to comprise a lower level of abstraction that might serve as a foundation for higher level symbolic thinking, which in turn may give rise to language and analogizing (Coolidge and Overmann 2012).
Researchers have presented stimuli at various levels of abstraction with categories ranging from concrete to more abstract. Concrete categories include exemplars that may look very much alike, such as bears versus humans, whereas abstract categories are broad and contain many diverse exemplars, such as “animal,” which contains insects, reptiles, fish, birds, and various mammals. Both great apes and American black bears have been show to form categories at each level of abstraction tested, when presented with natural categories, but it has proven difficult to determine which features these animals use to perform the tasks successfully (Vonk and Galvan 2014). The possibility remains that animals are relying on perceptual features, such as particular color, luminosity, and contrast patterns to control responding in these tasks, rather than generating an overarching category such as “animal.”
Perhaps the most abstract concept assessed in nonhumans is the capacity to represent mental states, in particular the mental states of others. This capacity is known as theory of mind (ToM) and has been widely studied in both human children and nonhuman species. Although evidence for ToM in nonhumans has been hotly contested (Penn and Povinelli 2007), researchers have begun to converge on the idea that at least some species, such as great apes, may have an understanding of some mental states, such as seeing and intentions. Evidence for understanding of knowledge states and beliefs is more equivocal. It has yet to be demonstrated that any nonhuman species represents the range and breadth of abstract concepts that can be communicated by humans through the use of symbolic systems, most notably language.
The extent to which nonhumans represent concepts for things that are abstractions of real objects remains uncertain as researchers struggle to make progress with experimental paradigms that will allow for a careful dissection of processes linked to observable features and those that depend upon an internal representation of an overarching conceptual category subsuming such features (Vonk and Povinelli 2006).
- Locke, J. (1975). An essay concerning human understanding. Oxford: Clarendon.Google Scholar
- Mackintosh, N. J. (2000). Abstraction and discrimination. In C. Heyes & L. Huber (Eds.), The evolution of cognition (pp. 123–141). Cambridge, MA: The MIT Press.Google Scholar
- Vonk, J. (2002). Can Orangutans and Gorillas acquire concepts for social relationships? International Journal of Comparative Cognition, 15, 257–277.Google Scholar
- Vonk, J., & Povinelli, D. J. (2006). Similarity and difference in the conceptual systems of primates: The unobservability hypothesis. In E. Wasserman & T. Zentall (Eds.), Oxford handbook of comparative cognition: Experimental explorations of animal intelligence (pp. 363–387). Oxford: Oxford University Press.Google Scholar