Theory of mind (ToM) refers to the more or less automatic tendency to impute mental states to oneself and others (Premack and Woodruff 1978). Among alternative terms such as folk psychology, naive psychology, mind reading and mentalizing (Astington and Baird 2005), ToM has emerged as the standard terminology among psychologists, primatologists, and philosophers. ToM is at the heart of human social cognition and underlies virtually every aspect of humanity.
FOLK PSYCHOLOGY AND APE PSYCHOLOGY
Consider, for example, how biological and physical motions are perceived. The motion of a falling apple is governed by Newton’s laws, and human infants as young as three months are sensitive to violations of such principles (Spelke 2000). In contrast, most human behaviors are interpreted in terms of desires, beliefs, and emotions. As a classic demonstration of how far we are ready to stretch our folk psychology, Fritz Heider and Marianne Simmel (1944) showed undergraduate students animation movies in which geometric figures moved in coordinated ways that strongly hinted intentional behaviors such as avoidance or helping. When asked to describe the scenes, participants did not hesitate to apply mentalistic attributes to circles and triangles.
Contemporary interest in ToM, however, was sparked by the mind-reading ability of Sarah, a fourteen-year-old chimpanzee who appeared to recognize intentions from behaviors. In a groundbreaking paper titled "Does the Chimpanzee Have a Theory of Mind?" (1978), David Premack and Guy Woodruff showed Sarah video vignettes of a trainer attempting in vain to achieve a goal (e.g., to exit a cage through a locked door) and then a pair of photographs depicting potential solutions (e.g., an intact key along with a broken one). Spontaneously, Sarah "recognized the videotape as problem, understood the actor’s purpose, and chose alternatives compatible with that purpose" (Premack and Woodruff 1978, p. 515). The authors attributed her success to the existence of a system that represents mental states, such as desire and intent, and links them to behaviors. They further maintained that "a system of inferences of this kind may properly be viewed as a theory because such states are not directly observable, and the system can be used to make predictions about the behavior of others" (p. 515).
Some thirty years later, the field remains divided over whether chimpanzees (or any nonhuman primates) possess a theory of mind or simply a theory of behaviors. Michael Tomasello and colleagues argue that the great apes understand others as "animate, goal-directed, and intentional agents" (2005, p. 675) even though they fall short in understanding beliefs and other mental states compared to humans (Call and Tomasello 1999). Others, however, caution that we have to first rule out the possibility that their responses are based on behavioral regularities (Heyes 1998; Povinelli and Vonk 2003, 2004). For example, chimpanzees spontaneously follow the gaze of others (Okamoto et al. 2002) and are attuned to what others can or cannot see (Hare et al. 2000; Povinelli and Eddy 1996). Nevertheless, it is unclear whether chimpanzees can link seeing to knowing. When competing for food, subordinate chimpanzees are reported to keep track of what dominant chimpanzees know and do not know about food locations (Hare et al. 2001). This is in stark contrast with an earlier finding that chimpanzees would beg for food indiscriminately from both trainers who could see and those who had a bucket over their head (Povinelli and Eddy 1996).
The current debate on ToM in nonhuman primates is not about whether they are capable of sophisticated social behaviors—yes, their abilities to keep track of and manipulate others sometimes rival those of humans. The issue is what can be concluded when their behaviors are compatible with both a ToM and a theory of behavioral regularities. Until a consensus is reached on this and other critical issues, the question Premack and Woodruff famously raised in their seminal work will remain unanswered.
FALSE BELIEFS AND THE DEVELOPMENT OF TOM IN HUMAN CHILDREN
In comparison, a consensus did emerge among developmental psychologists. It was suggested that an understanding of "false beliefs" is a sufficient condition for a ToM, because beliefs—unlike behaviors—are private and unob-servable, and thus are beyond the scope of a theory of behaviors (Dennett 1978). And thanks to language, beliefs can be identified unequivocally in human participants.
This idea was put to a test by Heinz Wimmer and Josef Perner (1983) when the researchers told young children a story in which Maxi put his chocolate in a kitchen drawer and left. In his absence, Maxi’s mom moved the candy to a different drawer. Children had to predict where Maxi would look for his chocolate on his return. While most six-year-old children correctly predicted Maxi’s action on the basis of his false belief that the candy was still in the original drawer, most four- and five-year-old children in this study expected Maxi to look in the new location.
Young children’s difficulties are not limited to thinking in other peoples’ shoes. They appear to be oblivious about false beliefs they held a few moments earlier (Gopnik and Astington 1988). The development of ToM is remarkably consistent and robust, progressing almost uniformly from appreciating that people have different desires to understanding false beliefs, and later to differentiating apparent versus true emotions (Wellman and Liu 2004). A meta-analysis of more than 170 false-belief studies concluded that the transition in false-belief understanding takes place in the preschool years for most children (Wellman et al. 2001).
Although passing the false-belief task attests to a ToM, the converse is not true (Bloom and German 2000). For example, younger children may fail due to their inability to inhibit improper responses (Leslie et al. 2004; Leslie and Polizzi 1998). In addition, children show a range of competences before they pass the false-belief test. At three months of age, infants prefer biological motion to random motion (Bertenthal et al. 1984) and shift their eye gaze after an adult makes an eye movement (Hood et al. 1998). At around twelve months, human infants begin to follow others’ line of sight and predict actions according to gaze (Sodian and Thoermer 2004). The second year of life sees the onset of joint visual attention (Butterworth and Jarrett 1991), imitation (Meltzoff 1995), and pretend play (Lillard 2002). There is even evidence that sixteen-month-old infants have an implicit understanding of false beliefs (Onishi and Baillargeon 2005). From three to five, children produce mental words such as think or want, start to appreciate jokes (Leekam 1991), and attempt to deceive (Sodian et al. 1991). ToM continues to develop; adults still fall victim to egocentrism from time to time (Keysar et al. 2003). Nonetheless, the basic apparatus for reading and reasoning about mind is largely in place early in childhood.
THEORIES OF THEORY OF MIND
What explains children’s initial failure and later success in the false-belief task? The theory theory postulates that we reason about mental events based on a system of heuristics or if-then rules (Gopnik and Wellman 1994). The child actively constructs and revises the informal ToM based on experience. For instance, a toddler may begin with a prediction that if Maxi wants something, he will act to satisfy his desire. Experiences with false beliefs of others or herself will force the child to abandon this simple heuristic and incorporate belief into an adultlike theory of mind (Bartsch and Wellman 1995).
The notion of a theory as the basis for mind reading is challenged by the simulation theory, which proposes that we compute others’ mental states through an automatic role-play simulation ("if I were him …"; Gallese and Goldman 1998; Gordon 1986). Young children find the impersonation difficult in a false-belief task because one has to withhold one’s own belief and instead reason with the erroneous belief of the person to be simulated.
If a child has to discover a ToM completely on his or her own, as suggested by the theory theory, the acquisition is unlikely to be consistent and universal. Two solutions have been proposed in the literature. One is the hypothesized theory ofmind mechanism (ToMM; Leslie et al. 2004; Baron-Cohen 1995), an innate neurological module that automatically maps behaviors of self-propelled agents to mental states and vice versa. The ToMM is thought to mature fairly early in life and kick off the subsequent development of folk psychology. Young children’s difficulty with the false-belief task is attributed to immature executive functioning, which works in conjunction with ToM to make behavioral predictions.
An alternative view traces the origin of ToM to language and culture. Everyday conversations expose children to the fact that speakers may have different knowledge and beliefs. Language also provides specific lexical terms for mental states (e.g., see, know, and want) and, in some cases, different syntactic structures for different mental states (e.g., "I want to …" but not "I want that…"; however "I know that…" but not "I know to…"; see de Villiers and de Villiers 2000). These linguistic structures invite children to think and talk about mental events in a way that is conventional in the language community. In this perspective, language and culture provide both the blueprint and building blocks for developing a ToM (Astington and Baird 2005).
MINDBLINDNESS AND BRAIN IMAGING
What would the world be without a ToM? Childhood autism is a spectrum of pervasive developmental disorders characterized by deficits in social interaction, verbal communication, and repetitive behaviors. Simon Baron-Cohen and colleagues (1985) hypothesized that autistic children have a severe deficit in the ToMM. For example, Baron-Cohen (1995) showed children a face ("Charlie") surrounded by four different kinds of candy, with Charlie’s eyes looking toward one of them. When asked "which candy does Charlie want," most four-year-olds pointed to where Charlie was gazing. In contrast, children with autism responded randomly to what Charlie wanted, even though they had no trouble answering where Charlie was looking. The problem is a failure to map perception to mental states. Niki L, a high-functioning autistic person, wrote:
Many [autistic persons] lack some sort of intuition and have a hard time guessing hidden rules many [normal] kids somehow see____ I think I eventually formed a relatively good theory of mind, but it took intentional effort. And I still have to apply it manually. It gets faster and faster as I collect many patterns in my memory, but I’m afraid it won’t be automatic forever. (Blackburn et al. 2000)
Brain imaging has begun to shed light on the potential neural mechanism underlying ToM. For example, Kevin Pelphrey and colleagues (2005) reported functional MRI evidence that participants with autism are insensitive to intentions conveyed by other people’s eye movements. Other imaging studies contrasted brain activities while processing mental versus physical events, for example, listening to stories involving rich mentalizing (Fletcher et al. 1995; Gallagher et al. 2000; Vogeley et al. 2001) or watching Heider/Simmel-like cartoon animations (Castelli et al. 2000). Consistently, stronger activations are found in the medial prefrontal cortex (MPFC), temporal poles, and the posterior superior temporal sulcus (STS) during mentalizing. For further information, see Frith and Frith (2003) and Baron-Cohen et al. (2000).
CONCLUSION
The prepositional phrase theory ofmindis a linguistic oddity. Contrary to the expectation of a native English speaker, for example, the "theory of mind mechanism" is not a theory about "mind mechanism" but a mechanism for theory of mind. The strong internal cohesiveness reflects a commitment to two of its core assumptions— that behaviors are understood in mental terms and that this understanding constitutes a theory.
Although both points are under vigorous debate, ToM as a field of scholarship has thrived, and continues to thrive, beyond the imagination of Premack and Woodruff (1978). Recent reports of self-recognizing elephants and dolphins (Plotnik et al. 2006; Reiss and Marino 2001) and deception in scrub jays (Dally et al. 2005), among others, question the view that ToM is unique to a handful of primate species—if not for humans only—and instead suggest similar competences have evolved independently as an adaptation to complex social lives. Meanwhile, computer scientists find ToM essential in developing autonomous, cooperative robotic agents (Scassellati 2002), and economists are tying many aspects of social exchange and decision making to ToM (Singer and Fehr 2005). This brief survey of ToM cannot do justice to a vast and vibrant field of study. But fortunately, motivated readers should not have difficulty finding in-depth and up-to-date readings that connect ToM with their specialty—after all, where there is social interaction, there is theory of mind. At least potentially.
