Most recent accounts of moral cognition have, explicitly or implicitly, been individualist theories of some kind. They fit with traditional approaches to cognitive science and psychology that place cognition firmly inside the head. The dominant methodological and metaphysical approach to studying cognition has been, and still is, individualism.1 Individualism is a methodological doctrine: cognition is explained through the study of the internal states of individuals. It is a metaphysical doctrine in that the individualist approach views cognitive phenomena as instantiated by the internal structure of individual agents, and by this internal structure alone. Thus the structure of the mind is an internal structure, and mental states and their structures are characterised and individuated independently of that which is external to those structures.
Individualism about psychology is the thesis that psychological states should be construed without reference to anything beyond the boundary of the individual who has those states. Put loosely, it is the view that for the purposes of scientifically understanding the mind, the individual is the boundary of cognition. (Wilson 2004:398)
Drawing the boundary of cognition in terms of the individual raises obvious issues with
1. Not to be confused with the individualism/anti-individualism debate over whether or not intentional mental states are individuated solely by the intrinsic properties of an individual or whether those states are individuated in terms of intrinsic properties and their relations to the external world. (Putnam 1975; Burge 1979).
respect to what constitutes the “individual”, and therefore where exactly the boundaries of cognition lie. Following the methodology of much of traditional cognitive science, I will take the bounds of cognition as typically posited by individualism in the cognitive sciences to be confined within the brain, or what Andy Clark refers to as being
“brainbound” (2008).2 Here we see an important contrast between the brain (and certain associated neural systems) and the non-neural body and external world. In more general terms, the human cognitive system is “confined to the head of the thinker” (Chemero and Silberstein 2008:3). Confining the cognitive system in this way means that the (non-neural) body and everything beyond is excluded from the cognitive system. This is not, of course, to claim that traditional accounts of cognition deny that one’s body and environment play important roles in, or that they are irrelevant to, the processes of cognition. But as we will see, these approaches attribute a very specific role to both the body and world, and the ways in which they interact with the mind – namely, they function as the inputs and outputs to the cognitive system but are not constitutive parts of high-level cognitive processes. For what is central to the individualist’s claim is that the actual processes of cognition itself (i.e. the proper parts of the cognitive system) are divorced from both the body and the environment:
The world is (just) a source of inputs and arena for outputs; the body is just an organ for receiving inputs and effecting outputs (actions); the task of early processing is to render the inputs as an inner world-model of sufficient thickness to allow the bulk of problem-solving activity to be defined over the inner model alone.
(Clark 2004:35)
To study cognition is to study what goes on in the head.
The dominant individualist views of cognition, since the advent of modern cognitive science, have been computational theories of mind, according to which cognition takes place via the inner manipulations of internal mental representations.3 There have been
2. Brainbound refers to the entire central nervous system (brain and spinal chord).
3. Note that not all individualist accounts of cognition are computational. For example, some have proposed that cognition is individualistic and best explained “in terms of large scale neural dynamics” as opposed to computation (see Chemero and Silberstein (2008) for a brief discussion). Nor are all computational accounts of cognition individualistic. But paradigmatic cases of computational theories of cognition typically are individualistic and vice versa. The point I will be pursuing in the following chapters is that what counts as a moral cognitive system (computational or otherwise) cannot be confined to an individualistic methodology nor metaphysics.
numerous accounts of how this computational architecture of the mind is actualised.
They include, but are in no way exhausted by, formal logical approaches, rule based systems, concept based systems, image based systems and various connectionist systems to name but a few (for a brief overview see Thagard 2008). Each proposes different accounts of what constitute representations and/or the systematic processes which act upon those representations. What they have in common is that they all converge on a particular explanation of how the mind works, namely through computational procedures operating upon internal representational structures: “thinking is essentially having and manipulating representations” (Bem and Looren de Jong 2006:280).
Take, for example, the two prominent individualistic accounts of cognition: the Computational Theory of Mind (most prominently associated with Jerry Fodor4 and associates (Fodor 1975; Pylyshyn 1984)) and connectionism (see for example Rumelhart and McClelland (1986)). Both approaches are computational. Cognition is best understood as computational algorithms acting upon representational structures (although some connectionist approaches claim to be neither computational nor representational). One of the main points of difference between these two accounts is what constitute the representational structures upon which the computations occur.
(Chemero and Silberstein 2008; Thagard 2005; Bem and Looren de Jong 2006).
The Fodorian style computationalist, for example, takes representations to be symbolic, formal language-like structures where cognising involves the manipulation of these structures. The claims of the Fodorian are heavily influenced by Chomskyian linguistics’
picture of an innate, species universal, generative grammar. Fodor, for example, holds that the mind has an innate, species universal, language-like mental architecture that accounts for the systematic structure and productivity of thought – the language of thought hypothesis (1975). (For similarities with the Principles and Parameters account of moral cognition, see section 3.2.1 below). Connectionism on the other hand views representations as activation patterns in neural networks. By applying particular rules of learning, the connectionist network is able to acquire exemplar or prototype based
4. To be fair to Fodor, he is at pains to explain that he does not think that a Computational Theory of Mind (CMT) accounts for all, or even most of our cognitive processes: “I certainly don’t suppose that it could comprise more than a fragment of a full and satisfactory cognitive psychology” (2000). This cannot be said for many other proponents of CTM, (especially those with an evolutionary psychological bent) whom Fodor targets in his book The Mind Doesn’t Work that Way: The Scope and Limits of Computational Psychology (2000).
conceptual representations. Cognitions (such as those involved in reasoning and inference making) operate by way of pattern completion processes acting over these representations. (See Churchland (1996) for suggestions as to how this type of approach might work in the moral domain.)
In each case, cognition consists in the computational processing (algorithmic manipulations or pattern activation) of representations (symbols or sub-symbolic activation patterns).5 What typifies these perspectives is that the mind possesses rich, internal (symbolic or sub-symbolic) structure, and the essence of cognition involves the manipulation of this structure. The model of the mind is that of an inner mental world populated by representations upon which the problem-solving activity is enacted.
Problems present in the environment are represented in the head. Cognition then solves the problem via the representational manipulations in the head, and produces the relevant output to which the body acts.
The computationalist theories of mind typically view the representational structures over which cognitions operate as being amodal – that is, separate from perception and motor systems which interface with the world. Perceptual inputs are transduced to amodal symbols over which cognitions occur, the output of which is transduced to motor response. In this way, modal sensory-motor systems (involving representational structures or otherwise) are seen as separate from high-level cognising, whereby the perceptual and motor systems provide the inputs and receive outputs from that cognitive process. Sensory-motor operations are peripheral to cognition proper. (See for example Pylyshyn 1984).
The above discussion has identified a number of positions (brainbound individualism, computationalism, and amodality), which need not be taken as a package deal. Where one draws the bounds of cognition need not commit one to any particular mental architecture nor to any particular view of the structure of representations. Moreover, how one views mental architecture need not commit one to any specific position on the bounds of cognition nor the modal nature of representations (if there are even any
5. There are debates in philosophy of mind as whether cognition is identical to computational processes, or whether cognition is realised by computational processes. For present purposes, I am neutral on these matters.
representations at all!). Taking all three together, however, does establish the traditional view of cognition: realised in the brain and involving computational processes acting over amodal representations. This tripartite account is what I will thus refer to as individualism.
3.1.1 A mental world of inner complexity
Individualistic accounts of cognition focus on representations and processes acting over those representations rather than physical interaction, and in doing so make a specific commitment to inner complexity. As I have outlined above, the individualist’s approach to understanding the human cognitive system is to look “inside”, for the cognitive system is bound within the brain and nervous system. Of central importance to individualistic explanations is internal structure and organisation. The most prominent individualistic accounts of cognition posit mental representations to account for this structure.
Mental representations are theoretical entities; they are posited by the sciences of the mind because they have rich explanatory value in that they account for the relationship between mind and world. They do this by doing what their name suggests, namely representing. Although accounts differ as to how representations actually go about representing the world, they are minimally construed as “causal surrogates for distal features of the environment” (Chemero and Silberstein 2008). In this sense, they bring aspects of the world inside, and in doing so allow cognitive processing about the world to occur in a manner that is independent or isolated from the world.6 Yet if this picture of the mind is correct, and our cognitive system is instantiated by the brain and neural systems, then it is not just the environment for which mental representations can be causal surrogates; they can also be causal surrogates for the non-neural body, its configurations and its movements.
As explained above, for computationalists such as the Fodorian and the connectionist this internal complexity is typically implemented by way of either symbolic or sub-symbolic structures. They often differ as to whether or not, and to what degree, these internal structures are innate or acquired. Fodor once famously (if not notoriously) held
6. Hence Clark dubs this form of inner world problem solving “isolationism” (Clark 2004).
the view that most of our concepts are innate (1975; 1981). On the other hand, connectionists working with neural networks often place much emphasis on the powerful associationist learning abilities of such networks. Although they differ somewhat as to the exact nature of the computational processes that act upon those structures, both accounts of the mind understand cognition as abstracting away from the physical and social environments in which cognitive processes are taking place (Wilson forthcoming; Marr 1982). If cognition is the proximate cause of behaviour, then this behaviour is due in part to the internal complexity of the cognitive system – be that complexity learned or otherwise.
What is explanatorily important from the individualist's perspective is that adaptive problems encountered in the world become problems to be solved inside the head.
External complexity of the problem domain becomes brainbound internal complexity.
We build complex inner models of complex outer worlds. To explain cognition and resultant behaviour is to look inwards.
3.1.2 The body and world
Since the individualist believes the cognitive system resides in the head, any explanations of cognition proper are themselves going to be confined to the head. Such reliance on internal models or representations in explanation reduces the importance and role of the body and environment (Haugeland 1998b). Not only does this make a specific commitment to the internal complexity of the mind, but in doing so these individualistic accounts relegate the environment and body as being external to the cognitive system.
The role of the body and world is to operate as the inputs and outputs to cognitive process, with the environment providing the adaptive problems to be solved and information for their solution (i.e., the inputs to the cognitive process). The body mediates those informational inputs (via perceptual apparatus), and enacts the action orientated outputs (via motor structures) to the environment. The environment is typically viewed as the problem domain and our perceptual apparatus as a “peripheral channel through which the problem is initially posed, and incidental facts are supplied”
(Haugeland 1998b:220).
Although the individualistic perspective does not deny that the environment and body both play important roles in cognition, it is clear that they do make a particular commitment to the nature of their role. The environment populates the contents of our minds over multiple timescales: the environment to which aspects of our minds are adapted, and those life-time environments within which cognition takes place. In each case, the individualist’s mind is furnished via some well defined, stable channels; be that through genetic channels such as inherited innate representational content (such as our ability to focus on faces soon after birth), or over one’s life history via perceptual inputs from the world around us. The body in this case provides a stable structure through which the mind can receive inputs from and interact with a variable world. The computational algorithms are constructed and operate with certain assumptions about the structure of the body and its properties that enable those algorithms and heuristics of the cognitive system to work reliably and successfully. In sum, we have environmental inputs via stable perceptual channels, cognition operating on representations of those inputs, and output through stable bodily structures with which to interact with the world.
So far we have seen that traditional accounts of cognition make specific commitments relating to the bounds and architecture of cognition. This view also makes specific commitments to the internal complexity of internal cognitive structures and the roles attributed to body and world. We will now look at the ways in which two prominent accounts of cognition are individualistic. Both have relevance to our larger explanatory project, the evolution of moral cognition.