The Instantiation of Higher Order Representations

Representational theories of consciousness come in different versions. One important position in the current debate is that consciousness is a kind of metarepresentation—either a higher-order perception or a higher-order thought. Such ideas have a long history in the philosophy of mind that dates back to John Locke, Gottfried Wilhelm Leibniz, Immanuel Kant, William James, and Franz Brentano. Important modern versions have been worked out by Armstrong (1980), Churchland (1985), Rosenthal (1986, 1990, 1993a, 1993b), and Metzinger (1993). Roughly speaking, to be conscious, in this view, is to be in a specific cognitive state, that of having a mental representation by which the system represents its own actual state as its own state. Being conscious of something means that a mental representation is accompanied by, or embedded in, a second, higher-order representation to the effect that the system itself is in a certain state. The attractive feature of such concepts is that they could be a point of departure for the development of concrete hypotheses on the realization of such states in brains and possibly in other systems as well.

The higher-order representation (HOR) concept implies that phenomenal states cannot be understood in isolation. It resembles the concept that Kant introduced in the context of rejecting Hume's associationism. Phenomenal experience is not merely a succession or a "heap" of different cognitive states, but an integrated experience of a world of objects and of ourselves as subjects within it (Strawson 1966; van Gulick 1993). Phenomenal states are cognitive events that presuppose a model of the world and of the self, to which first-order representations can be bound. Information from different modalities and memory stores must be joined to one coherent representational structure.

If such representational structures are instantiated in a neural net, it will necessarily require the binding of segregated, widely distributed data into one Hebbian assembly. Thus, one consequence of the HOR concept would be to postulate the existence of large-scale neuronal assemblies and also a specific binding mechanism that makes such assemblies possible. The core of my hypothesis is that the NMDA synapse implements the binding mechanism that the brain uses to produce these large-scale representations to which HORs belong (Flohr 1991).

Baars (1988) proposed that one of the computational functions of consciousness is to broadcast information throughout the central nervous system. According to the present hypothesis, a rich set of activated associative links and widely available information is both the precondition for the occurrence of conscious states and the reason for the specific computational role they have. For inner states Kant assumed a similar necessary interdependence between their connectedness, consciousness, and functional role. According to Kant's concept of synthesis, inner states must at least be capable of being connected with all other cognitive states in one consciousness to acquire the functional status of a mental representation (Kant 1781, A116; Strawson 1966; Kitcher 1990).

Was this article helpful?

0 0

Post a comment