Carnegie Mellon


Infant Cognition Lab

Research Interests


Publications & Papers


Carnegie Mellon Links:
Academic Departments
Administrative Departments
Site Index

Search Carnegie Mellon:



Research Interests

My general area of research is cognitive development in infancy and early childhood. I focus on the early development of categorization and the development of the animate-inanimate distinction, both of which are among the most fundamental cognitive skills.  Categorization is especially important to infants, young children, and adults as it is the primary means of coding experience, which in turn reduces demands on inherently limited memory storage and perceptual and reasoning processes. The development of the concept of animacy represents the most basic division between different ontological kinds, and it is thought to be a crucial building block for children’s emerging representations about the world around them. In my work, I am attempting to provide an account for the development of conceptual knowledge in infancy and childhood, with a focus on the relationship between early categorization and knowledge about natural kinds and artifacts. Thus far, I have shown that infant rely on perceptual features and the functions of those features to categorize objects, and that learning the associations among these properties in turn may lead to the development of deeper, conceptual knowledge about ontological kinds.

Research on early categorization

My primary research focus has been on the development of categorization in infancy. In much of this research, I have used the sequential touching or object manipulation technique in which infants are presented with eight toys from two categories – say, for example, four animals and four vehicles - and allowed to play with them freely. If infants know or detect that some of the objects are alike, they will touch sequentially those objects. Using this technique, George Butterworth and I (Rakison & Butterworth, 1998a, 1998b), found that infants, particularly those under 22 months, do not have a conceptual understanding of category members as “the same kind of thing”, as has been generally assumed. Rather, infants rely on the perceptible properties of objects to categorize superordinate domains like animals and vehicles. In particular, infants attend to the parts – for example, legs, wheels - of objects and the structural configuration given by those parts.

In a later series of experiments with the same paradigm, I found infants categorize at the basic level by attending to parts of objects and not by using knowledge about category relatedness or nonobvious properties (Rakison & Cohen, 1999). Infants attended only to those parts (e.g., legs, wings) with a functional significance for an object (e.g., movement type) and not to smaller, functionally insignificant parts (e.g., facial features).

Research on the development of an animate-inanimate distinction:

A second program of research concerns the development of infants’ notion of animacy, the acquisition of which is thought to be one of the cornerstones of cognitive development (e.g., Leslie, 1995; Mandler, 1992). This line of study is particularly crucial because very little is known about how infants begin to categorize and make inductive inferences on the basis of objects’ internal, nonobvious attributes rather than their external, perceptual appearance. I am working on the mechanisms that underlie the development of infants’ understanding that distinct ontological kinds possess different physical and psychological causal characteristics. Specifically, I am examining whether 10- to 18-month-old infants acquire different aspects of the distinction between animate and inanimate objects by attending to object properties such as type of motion, shape, and functional parts.

In one series of studies with the habituation procedure, I tested the hypothesis that emerged from my earlier work that infants’ earliest distinction between animates and inanimates comes from the correlation between large, moving, functional parts and the path of motion of an object (e.g., linear versus nonlinear). The results supported the hypothesis, and suggest a developmental trend whereby 14-month-old infants associate object motion to relevant object parts and then later, around 18 months, extend this correlation to whole objects (Rakison & Poulin-Dubois, under review).
Click here to see a movie of the stimuli and test yourself! (requires Shockwave).. In addition, I am examining whether part-motion correlations provide the foundation for infants’ knowledge of animates as agents in causal events and as self-propelling entities.

I have also examined infants’ inductions about the movements of objects. Using the inductive generalization procedure developed by Mandler and McDonough (1996), I have tested whether infants generalize motion events from one category member to another. The data suggest that infants’ understanding of the forms of motion of different domains develops later than previously thought, and is based on perceptual attributes such as shape and parts. Moreover, I have found that infants’ performance is better described in terms of imitation and perceptual matching than as induction based on conceptual knowledge (Poulin-Dubois & Rakison, under review).

Research on Predator Avoidance Mechanims in Infants and Adults: click here.