Evolving Self-Consciousness - Peter Carruthers, Logan Fletcher, J. Brendan Ritchie

This is one of the sections from the Online Consciousness Conference (2012) that I have found intriguing and thought-provoking..

Special Session on the Developmental Conditions of Self Consciousness organized by James Dow, Hendrix College

1. James Dow, Hendrix College On the Developmental Conditions of Self-Consciousness

2. Peter Carruthers, Logan Fletcher, and J. Brendan Ritchie (all at) University of Maryland Evolving Self-Consciousness

   Commentators: Joel Smith, University of Manchester JeeLoo Liu, California State University, Fullerton

3. Radu J. Bogdan, Tulane University
   Self-Consciousness: Executive Design, Sociocultural Grounds

   Commentators: Kyle Ferguson Graduate Center, CUNY Robert Lurz, Brooklyn College, CUNY Henry Shevlin Graduate Center, CUNY

Particularly, it was the article by Peter Carruthers, Logan Fletcher, and J. Brendan Ritchie, Evolving Self-Consciousness, that I found interesting (the article from James Dow, On the Developmental Conditions of Self-Consciousness, is also interesting).

The authors offer two differing accounts of how we have become capable of being self-conscious, and conclude that one - a third-person account based in mind-reading - is more likely the the other - a first-person based need for metacognitive control and awareness. Integrally, the answer is that both are true and interdependent, likely with other mechanisms also playing a role.

Evolving Self-Consciousness
Peter Carruthers, Logan Fletcher, and J. Brendan Ritchie
University of Maryland

Humans have the capacity for awareness of many aspects of their own mental lives—their own experiences, feelings, judgments, desires, and decisions. We can often know what it is that we see, hear, feel, judge, want, or decide. This article examines the evolutionary origins of this form of self-consciousness. Two alternatives are contrasted and compared with the available evidence. One is first-person based: self-consciousness is an adaptation designed initially for metacognitive monitoring and control. The other is third-person based: self-consciousness depends on the prior evolution of a mind-reading system which can then be directed toward the self. It is shown that the latter account is currently the best supported of the two.

1. Introduction
There are a number of kinds of self-consciousness. One is awareness of oneself as a bodily agent, as established by the so-called “mirror test” (Gallup, 1970). While interesting, this form of self-consciousness has little to do with awareness of oneself as a cognitive being. Rather, the mirror test measures an ability to notice cross-modal contingencies, becoming aware of the mapping between one’s own bodily movements (as experienced proprioceptively) and what one perceives in the mirror. Another—much more demanding—form of self-consciousness concerns awareness of oneself as an on-going bearer of mental states and dispositions, who has both a past and a future. In effect, this form of self-consciousness seems to require a conception of oneself as a self, together with a capacity for narrative, weaving one’s current thoughts and experiences into a larger story of one’s life.

Situated somewhere between these two—more demanding than agentive self-awareness but less demanding than awareness of oneself as an ongoing self—is the form of self-consciousness that is the focus of this article. This is awareness of one’s own current mental states: one’s judgments, beliefs, desires, values, decisions, intentions, experiences, and emotions. Humans undoubtedly enjoy such self-awareness. We don’t just see, we are aware that we see; we don’t just hear, we are aware that we hear; and so on. And we often know what we think, want, decide, or fear. Our question concerns the evolutionary roots of these capacities for self-knowledge.

This paper will assume that capacities for self-consciousness are rooted in some kind of distinct adaption in addition to general learning abilities. This assumption is not uncontroversial. Some might be tempted to endorse empiricism about concepts and concept acquisition, for example (Prinz, 2002), while claiming that the classifications that we make among our own mental states and the knowledge that we have of their patterns of interaction and contributions to behavior are a product of general learning (whether associative, or involving some sort of inference to the best explanation, or both). This account strikes us as quite implausible. But for present purposes we will simply assume, without argument, that it is false.

One can then envisage two broad accounts of the evolution of a capacity for self-consciousness. One is first-person based. It is that self-consciousness evolved for purposes of metacognitive monitoring and control. On this account, organisms evolve a capacity for self-consciousness in order better to manage and control their own mental lives. By being aware of some of their mental states and processes, organisms can become more efficient and reliable cognizers, and can make better and more adaptive decisions as a result.(1)

The first-person-based view is consistent with a range of accounts of the cognitive capacities or mechanisms underlying self-consciousness. At one extreme are those who believe in mechanisms of so-called “inner sense” (Nichols and Stich, 2003; Goldman, 2006). Just as our regular senses detect, and enable us to have knowledge of, properties of the external world and of our own bodies, so inner sense is supposed to enable us to detect and have knowledge of our own mental lives. At the other extreme one might postulate just a body of core knowledge, similar to the knowledge proposed in the domains of physics and number (Spelke and Kinzler, 2007). This would contain a set of representational primitives like THINKS and WANTS, together with some basic inferential principles that would enable one to predict the impact of some simple self-directed interventions. The executive systems that deploy this knowledge would have access to just the same “globally broadcast” perceptual and imagistic information as do other decision-making systems, and would lack any special channels of access to the subject’s own non-sensory mental states.

The first-person-based view is also consistent with a range of accounts of the relationship between self-consciousness and third-person mind-reading. On one view, it might be claimed that the mechanisms of inner sense are exapted and used when simulating the minds of others, in such a way that capacities for mind-reading depend upon our capacity for self-consciousness (Goldman, 2006). Likewise it might be claimed that the core knowledge that underlies self-consciousness is re-deployed (either by evolution or by individual learning) to provide the basis for third-person mind-reading. Alternatively, it might be claimed that capacities for self-consciousness and for mind-reading are independent of one another (Nichols and Stich, 2003).

Since theories are stronger (less open to attack) that make fewer assumptions, our focus in this article will be on a minimalist “core knowledge” first-person-based account of the adaptive basis of self-consciousness, which makes no claim to explain the basis of mind-reading. Hence the first-person-based account to be considered here holds that self-consciousness and mind-reading are independent capacities. Moreover, the account of self-knowledge in play is consistent with the “interpretive sensory-access” (ISA) theory defended by Carruthers (2011), and is not directly targeted by the critiques of other views that are mounted in that work. Because these assumptions are significantly more minimal than any that are made in the existing literature, if they turn out to be indefensible then by the same token all existing first-person-based views will also be undermined.

The contrasting account of the evolution of self-consciousness is third-person based. It maintains that the adaptation underlying the capacity for knowledge of one’s own mental states is a mind-reading faculty (consisting of a body of core knowledge about the mind, or a domain-specific learning mechanism with representational primitives, or both), which evolved initially for social purposes (Carruthers, 2011). These purposes might be competitive, as “Machiavellian intelligence” accounts of the evolution of mind-reading maintain (Byrne and Whiten, 1988, 1997), or cooperative (Richerson and Boyd, 2005; Hrdy, 2009), or both. The mind-reading faculty would have access to globally broadcast perceptual and imagistic representations as input, and attributions of mental states to oneself would initially utilize this input together with the same core knowledge and principles that are employed for third-person mind-reading. (Some first-person principles might subsequently be learned, of course.) In effect, self-consciousness results from turning our evolved mind-reading capacities on ourselves.(2)

In what follows we will compare the empirical predictions made by these first-person-based and third-person-based accounts, and confront them with the available data. Section 2 will focus on the expected signature effects of the adaptations that these theories postulate, before Section 3 turns to evidence from comparative psychology.(3)