Weiskopf on Cognitive Systems 4

When the Terminator retrieves information from memory, it appears as text in his visual field. Presumably this text is read by him and used to guide his murderous actions. But this is a purely internal process, and it is plausibly cognitive. (Weiskopf, 2010)

I don't find it that plausible.  To me, this is just Andy's Martian with the bitmap "mental" image again.

Weiskopf on Cognitive Systems 3

The criterion is this: the boundaries of a cognitive system are given by the location of its transducers and its effectors. A transducer, in Pylyshyn’s terms (pp. 151–178) is a device that (1) maps inputs described in physical terms into outputs described in representational terms in a way that is (2) interrupt-driven and (3) primitive and nonsymbolic. Saying that transducers are interrupt-driven is just to say that their activation is mandatorily determined by the presence of their physical input conditions. Saying that they are primitive implies that they do not carry out their mapping function by any internal representational means; their operations do not involve cognitive processes, although they may obviously be physically complex. The most important condition on transducers, for our purposes, is that they have the function of turning physical stimuli into representational or computational  states. The inputs to a transducer are not themselves representational; transducers respond only to physical properties and magnitudes. They take, for example, pressure, temperature, vibrations in the air, or ambient light in a region of space, and produce vehicles that  represent something, most frequently some aspect of the environment that the stimulus typically carries information about. Transducers can thus be thought of as the place in where things in the external environment become input for the cognitive system.
     The same can be said of effectors. Corresponding to the above definition of a transducer, an effector is a device that (1) maps inputs described in representational terms into outputs described in physical terms in a way that is (2) interrupt-driven and (3) primitive and nonsymbolic. That is, an effector does what a transducer does, but in reverse. It takes a representation and produces a physical event; for example, activation pattern in certain muscle groups. The input representation can be understood as something like a direct motor command, and this command acts immediately on the body. Both transducers and effectors are important for delimiting systems, but for brevity I will sometimes simply call this the transducer view of systems.  (Weiskopf, 2010)

So, is a single neuron a cognitive system by this criterion?  Sounds like yes to me.

Weiskopf on Cognitive Systems 2

The criterion is this: the boundaries of a cognitive system are given by the location of its transducers and its effectors. A transducer, in Pylyshyn’s terms (pp. 151-178) is a device that (1) maps inputs described in physical terms into outputs described in representational terms in a way that is (2) interrupt-driven and (3) primitive and nonsymbolic. (Weiskopf, 2010)

 So, a being without sense organs that mentally computes a sequence of Fibonacci numbers would not be a cognitive agent?

Weiskopf on Cognitive Systems

A cognitive system is a set of physical structures and mechanisms that collectively realize a specific functional architecture.  Such architecture makes available a representational vocabulary, a set of primitive operations defined over them, a set of resources that these operations may make use of, and a set of control structures that determine how the activation and inhibition of operations and resources is orchestrated. These collectively determine the internal dynamics of processes in the system: how one set of input representations triggers a cascade of processing throughout various parts of the system, resulting eventually in some
sort of output.  (Weiskopf, 2010, p. ??)

So, that seems to me to be a reasonable way to demarcate a cognitive system, but it's not exactly what Weiskopf goes for.  More on this later.

Weiskopf Review of Menary's Cognitive Integration

This has apparently been out for a bit, but here it is.
Mind, Volume119, Issue 474, pp. 515-519

Weiskopf on Embodied Cognition

There is what I think is a very nice paper by Weiskopf dealing with embodied cognition in a slightly out of the way place, Studies in History and Philosophy of Science.   Additionally, there is a reply to Weiskopf by Raymond Gibbs and Marcus Perlman and a rejoinder by Weiskopf.  I have not read the latter two, but I think it should be a useful exchange.

I remember the first time I heard Dan give this paper, I was sitting with Fred Adams, and Fred kept saying, "That's what I'm gonna say!"  So, there are commonalities between Fred's paper that has just appeared in Phenomenology and the Cognitive Sciences and Dan's. 

Coincidentally, the volume of SHPS was edited by someone familiar to the EC crowd, Mark Sprevak.  It is a special issue on Computation and Cognitive Science.  That guy Aizawa has a paper in the collection as well.

New Paper by Weiskopf

I think Dan does a very nice job in "Embodied cognition and linguistic comprehension".  I'm basing this on a version I saw a few years back.

Weiskopf Review of Menary's Cognitive Integration

In Mind.