5.10: The Architecture of Mind Reading

Social interactions involve coordinating the activities of two or more agents. Even something as basic as a conversation between two people is highly coordinated, with voices, gestures, and facial expressions used to orchestrate joint actions (Clark, 1996). Fundamental to coordinating such social interactions is our ability to predict the actions, interest, and emotions of others. Generically, the study of the ability to make such predictions is called the study of theory of mind, because many theorists argue that these predictions are rooted in our assumption that others, like us, have minds or mental states. As a result, researchers call our ability to foretell others’ actions mind reading or mentalizing (Goldman, 2006). “Having a mental state and representing another individual as having such a state are entirely different matters. The latter activity, mentalizing or mind reading, is a second-order activity: It is mind thinking about minds” (p. 3).

There are three general, competing theories about how humans perform mind reading (Goldman, 2006). The first is rationality theory, a version of which was introduced in Chapter 3 in the form of the intentional stance (Dennett, 1987). According to rationality theory, mind reading is accomplished via the ascription of contents to the putative mental states of others. In addition, we assume that other agents are rational. As a result, future behaviors are predicted by inferring what future behaviors follow rationally from the ascribed contents. For instance, if we ascribe to someone the belief that piano playing can only be improved by practicing daily, and we also ascribe to them the desire to improve at piano, then according to rationality theory it would be natural to predict that they would practice piano daily.

A second account of mentalizing is called theory-theory (Goldman, 2006). Theory-theory emerged from studies of the development of theory of mind (Gopnik & Wellman, 1992; Wellman, 1990) as well as from research on cognitive development in general (Gopnik & Meltzoff, 1997; Gopnik, Meltzoff, & Kuhl, 1999). Theory-theory is the position that our understanding of the world, including our understanding of other people in it, is guided by naive theories (Goldman, 2006). These theories are similar in form to the theories employed by scientists, because a naive theory of the world will—eventually—be revised in light of conflicting evidence.

Babies and scientists share the same basic cognitive machinery. They have similar programs, and they reprogram themselves in the same way. They formulate theories, make and test predictions, seek explanations, do experiments, and revise what they know in the light of new evidence. (Gopnik, Meltzoff, & Kuhl, 1999, p. 161)

There is no special role for a principle of rationality in theory-theory, which distinguishes it from rationality theory (Goldman, 2006). However, it is clear that both of these approaches to mentalizing are strikingly classical in nature. This is because both rely on representations. One senses the social environment, then thinks (by applying rationality or by using a naïve theory), and then finally predicts future actions of others. A third theory of mind reading, simulation theory, has emerged as a rival to theory-theory, and some of its versions posit an embodied account of mentalizing.

Simulation theory is the view that people mind read by replicating or emulating the states of others (Goldman, 2006). In simulation theory, “mindreading includes a crucial role for putting oneself in others’ shoes. It may even be part of the brain’s design to generate mental states that match, or resonate with, states of people one is observing” (p. 4).

The modern origins of simulation theory rest in two philosophical papers from the 1980s, one by Gordon (1986) and one by Heal (1986). Gordon (1986) noted that the starting point for explaining how we predict the behavior of others should be investigating our ability to predict our own actions. We can do so with exceedingly high accuracy because “our declarations of immediate intention are causally tied to some actual precursor of behavior: perhaps tapping into the brain’s updated behavioral ‘plans’ or into ‘executive commands’ that are about to guide the relevant motor sequences” (p. 159).

For Gordon (1986), our ability to accurately predict our own behavior was a kind of practical reasoning. He proceeded to argue that such reasoning could also be used in attempts to predict others. We could predict others, or predict our own future behavior in hypothetical situations, by simulating practical reasoning.

To simulate the appropriate practical reasoning I can engage in a kind of pretend-play: pretend that the indicated conditions actually obtain, with all other conditions remaining (so far as is logically possible and physically probable) as they presently stand; then continuing the make-believe try to ’make up my mind’ what to do given these (modified) conditions. (Gordon, 1986, p. 160)

A key element of such “pretend play” is that behavioral output is taken offline.

Gordon’s proposal causes simulation theory to depart from the other two theories of mind reading by reducing its reliance on ascribed mental contents. For Gordon (1986, p. 162), when someone simulates practical reasoning to make predictions about someone else, “they are ‘putting themselves in the other’s shoes’ in one sense of that expression: that is, they project themselves into the other’s situation, but without any attempt to project themselves into, as we say, the other’s ‘mind.’” Heal (1986) proposed a similar approach, which she called replication.

A number of different variations of simulation theory have emerged (Davies & Stone, 1995a, 1995b), making a definitive statement of its fundamental characteristics problematic (Heal, 1996). Some versions of simulation theory remain very classical in nature. For instance, simulation could proceed by setting the values of a number of variables to define a situation of interest. These values could then be provided to a classical reasoning system, which would use these represented values to make plausible predictions.

Suppose I am interested in predicting someone’s action. . . . I place myself in what I take to be his initial state by imagining the world as it would appear from his point of view and I then deliberate, reason and reflect to see what decision emerges. (Heal, 1996, p. 137)

Some critics of simulation theory argue that it is just as Cartesian as other mind reading theories (Gallagher, 2005). For instance, Heal’s (1986) notion of replication exploits shared mental abilities. For her, mind reading requires only the assumption that others “are like me in being thinkers, that they possess the same fundamental cognitive capacities and propensities that I do” (p. 137).

However, other versions of simulation theory are far less Cartesian or classical in nature. Gordon (1986, pp. 17–18) illustrated such a theory with an example from Edgar Allen Poe’s The Purloined Letter:

When I wish to find out how wise, or how stupid, or how good, or how wicked is any one, or what are his thoughts at the moment, I fashion the expression of my face, as accurately as possible, in accordance with the expression of his, and then wait to see what thoughts or sentiments arise in my mind or heart, as if to match or correspond with the expression. (Gordon, 1986, pp. 17–18)

In Poe’s example, mind reading occurs not by using our reasoning mechanisms to take another’s place, but instead by exploiting the fact that we share similar bodies. Songwriter David Byrne (1980) takes a related position in Seen and Not Seen, in which he envisions the implications of people being able to mold their appearance according to some ideal: “they imagined that their personality would be forced to change to fit the new appearance. . . .This is why first impressions are often correct.” Social cognitive neuroscience transforms such views from art into scientific theory.

Ultimately, subjective experience is a biological data format, a highly specific mode of presenting about the world, and the Ego is merely a complex physical event—an activation pattern in your central nervous system. (Metzinger, 208, p. 208)

Philosopher Robert Gordon’s version of simulation theory (Gordon, 1986, 1992, 1995, 1999, 2005a, 2005b, 2007, 2008) provides an example of a radically embodied theory of mind reading. Gordon (2008, p. 220) could “see no reason to hold on to the assumption that our psychological competence is chiefly dependent on the application of concepts of mental states.” This is because his simulation theory exploited the body in exactly the same way that Brooks’ (1999) behavior-based robots exploited the world: as a replacement for representation (Gordon, 1999). “One’s own behavior control system is employed as a manipulable model of other such systems. . . . Because one human behavior control system is being used to model others, general information about such systems is unnecessary” (p. 765).

What kind of evidence exists to support a more embodied or less Cartesian simulation theory? Researchers have argued that simulation theory is supported by the discovery of the brain mechanisms of interest to social cognitive neuroscience (Lieberman, 2007). In particular, it has been argued that mirror neurons provide the neural substrate that instantiates simulation theory (Gallese & Goldman, 1998): “[Mirror neuron] activity seems to be nature’s way of getting the observer into the same ‘mental shoes’ as the target—exactly what the conjectured simulation heuristic aims to do” (p. 497–498).

Importantly, the combination of the mirror system and simulation theory implies that the “mental shoes” involved in mind reading are not symbolic representations. They are instead motor representations; they are actions-on-objects as instantiated by the mirror system. This has huge implications for theories of social interactions, minds, and selves:

Few great social philosophers of the past would have thought that social understanding had anything to do with the pre-motor cortex, and that ‘motor ideas’ would play such a central role in the emergence of social understanding. Who could have expected that shared thought would depend upon shared ‘motor representations’? (Metzinger, 2009, p. 171)

If motor representations are the basis of social interactions, then simulation theory becomes an account of mind reading that stands as a reaction against classical, representational theories. Mirror neuron explanations of simulation theory replace sense-think-act cycles with sense-act reflexes in much the same way as was the case in behavior-based robotics. Such a revolutionary position is becoming commonplace for neuroscientists who study the mirror system (Metzinger, 2009).

Neuroscientist Vittorio Gallese, one of the discoverers of mirror neurons, provides an example of this radical position:

Social cognition is not only social metacognition, that is, explicitly thinking about the contents of some else’s mind by means of abstract representations. We can certainly explain the behavior of others by using our complex and sophisticated mentalizing ability. My point is that most of the time in our daily social interactions, we do not need to do this. We have a much more direct access to the experiential world of the other. This dimension of social cognition is embodied, in that it mediates between our multimodal experiential knowledge of our own lived body and the way we experience others. (Metzinger, 2009, p. 177)

Cartesian philosophy was based upon an extraordinary act of skepticism (Descartes, 1996). In his search for truth, Descartes believed that he could not rely on his knowledge of the world, or even of his own body, because such knowledge could be illusory.

I shall think that the sky, the air, the earth, colors, shapes, sounds, and all external things are merely the delusions of dreams which he [a malicious demon] has devised to ensnare my judgment. I shall consider myself as not having hands or eyes, or flesh, or blood or senses, but as falsely believing that I have all these things. (Descartes, 1996, p. 23)

The disembodied Cartesian mind is founded on the myth of the external world.

Embodied theories of mind invert Cartesian skepticism. The body and the world are taken as fundamental; it is the mind or the holistic self that has become the myth. However, some have argued that our notion of a holistic internal self is illusory (Clark, 2003; Dennett, 1991, 2005; Metzinger, 2009; Minsky, 1985, 2006; Varela, Thompson, & Rosch, 1991). “We are, in short, in the grip of a seductive but quite untenable illusion: the illusion that the mechanisms of mind and self can ultimately unfold only on some privileged stage marked out by the good old-fashioned skin-bag” (Clark, 2003, p. 27).