One amazing ability that very young children seem to have is that of marking a certain category of acts as pretend, not to be taken seriously. Natural selection would certainly opt for a cognitive system that constructs models of how the world actually is. How is it that this system, especially very young versions of it, can purposefully misconstrue the world? Furthermore, given that adults most often present to the child the world as it actually is, how do children know when to make exception, and construe an adult's (or anyone else's) behavior as pretense instead?
How does the young child come to understand that these
adults are pretending to sing into microphones--that the
bananas are being used in a symbolic way?
What enables children to make a pretense interpretation is a largely unexplored question about one of childhood's major activities, and our research aims to make important headway towards discovering the conditions that accompany a pretense interpretation. Our hypothesis is that pretense acts are characterized by specific behavioral regularities. Such regularities may serve as cues to pretense interpreters, effectively signaling, "This is pretense."

To examine this, we have been analyzing matched pretense and real episodes for behavioral regularities that are uniquely associated with the pretense mode. In a first experiment, parents were asked to pretend to have a snack, and to really have a snack, while seated across a table from their 18-month-olds. Time-synchronized hidden videocameras filmed both parent and child. In some experiments, a motion analyzing system measures the parents' movements during both events, and in others, a the Computer Speech Lab is used to analyze features of parents' voices that vary across settings. The pretend and real snack events are compared for the relative frequency and degree of several possible cues.

Pilot and animal work suggested that several behaviors may consistently differ, such as facial configurations (smiles, raised brows); gestures that are oddly timed, or that exceed or fall short of their "real" paths of motion; and the use of sounds that simulate real counterparts (like saying "gulp" when pretending to drink). Many have commented that pretending involves "knowing smiles," but we know of no descriptions of exactly what such a smile (if it exists) looks like.

Dressed for Halloween as "the sea," this young child
extends the pretense by going fishing.
Using the Facial Action Coding system (Ekman & Rosenberg, 1997), we will examine whether the smiles parents emit during pretense events differ from those issued during real events, and if so, how. Other experiments are ongoing to examine the generalizability of these regularities to home settings, to nonpretense and to other pretense acts, and to addressees of different ages. We are also examining links to other early social cognitive skills, like social referencing and joint attention.

A second major question we will address is which of those behaviors that are regularly provided with pretense acts actually indicate to observers that those acts are pretense. This is being done by showing the episodes filmed in the first experiment to children and adults, to ensure they interpret the acts as pretense or real. In addition, we have been systematically filtering cues. For example, to some participants we have played only the audio portion, or shown only the face, and in each case participants have judged whether the events are pretense or real. In another set of experiments, we will examine whether children's ability to interpret pretense improves when the signs of pretense are provided.

This research is providing an important first step in a rich program to decipher how pretense acts are conveyed and understood. The work is novel in its focus on objective measurement of paralinguistic cues to pretense, its direct comparison of pretense and real acts, and its examination of how parents pretend in front of young children. Paralinguistic cues to pretense have been anecdotally mentioned in the literature but have rarely been subjected to objective and programmatic measurement and analysis.

Back to Signs of Pretense