RT list: Is there a spreading activation network inside a conceptual address?

From: Dan Sperber (dan@sperber.com)
Date: Tue Jan 24 2006 - 13:03:43 GMT

  • Next message: psousa@umich.edu: "RT list: New Email Address"

    Luis asks an interesting question to which RT
    does not, and is not meant to, provide a direct
    answer. The study of memory is a well-developed
    field within cognitive psychology and
    neuroscience and what we should aim at is
    fruitful interaction with it. This goes both
    ways. Whatever is found about memory is relevant
    to us because it determines accessibility which
    affects relevance. Conversely, the Cognitive
    Principle (human cognition is geared towards the
    maximization of relevance) has implications for
    memory: we should in particular expect memory to
    be so organised as to favor the activation of
    what is likely to be relevant to ongoing
    cognitive processes. Chunking in particular (and
    a concept is chunk - which may itself have
    sub-parts) should reflect probabilities of
    various pieces of information being relevant in
    the same contexts. This in turn is likely to be
    related to objective properties of the world and
    of its affordances for humans. The idea of
    spreading activation is one that fits
    particularly well with ideas in RT, since it
    allows for the effect of various contextual
    factors of relevance to interact and to give
    greater accessibility to pieces of informations
    linked to more factors. Luis raises a more
    specific question: do we need two levels of
    networking: one intra-conceptual, the other
    inter-concepts, so to speak. This seems too rigid
    at first blush, if only because much of the
    information within a conceptual entry links it to
    other concepts, so how would you keep levels
    separate? In fact, one classical way to look at
    conceptual entries is to think of them as the
    network of links of a given node. This
    alternative by itself might be too loose. Imagine
    as a third, more attractive alternative a single
    network with differences in thresholds of
    activation defining distinct areas and sub-areas
    corresponding to concepts, for instance. But the
    right description is for the (neuro-)psychology of memory to discover.

    Cheers, Dan

    At 06:19 24/01/2006, you wrote:

    >Hello, a question:
    >I understand that in the RT framework some form
    >of a spreading activation model of memory is
    >accepted, so that nodes can be said to be linked
    >with other nodes containing information in an
    >intricate complex system. I also understand that
    >three types of information can be stored in a
    >conceptual address or node: the logical entry,
    >the encyclopedia entry, and the lexical entry.
    >However, my question pertains specifically to
    >the information stored intrinsically in a
    >particular conceptual address or node. If a
    >spreading activation model is used to link nodes
    >and other information together, then can it also
    >be said that there is another deeper level
    >network of spreading activation taking place
    >intrinsically in a particular node which links
    >information from a logical entry, encyclopedic
    >entry, and lexical entry? Perhaps due to the
    >inferential nature of the logical entry it may
    >not be linked to a network at all (I don’t
    >know). At any rate, to be more precise, what I
    >am really pondering is if the information in the
    >encyclopedic entry (a vast store of potential
    >information) is also stored and organized
    >intrinsically in a particular node in some form
    >of a spreading activation network? Are there
    >actually two layers of spreading activation
    >networks at work inside the mind? Is there a
    >spreading activation network linking the nodes,
    >and another (perhaps more complicated) spreading
    >activation network linking the information of
    >the encyclopedic entry specifically inside a
    >node? This may be something well known to
    >others, but I am not familiar with this. Perhaps
    >someone can direct me to any psycholinguistic research addressing this issue.
    >Does anyone have any comment on this?
    >Greatly appreciated,
    >-Luis C. Reyes

    This archive was generated by hypermail 2b29 : Tue Jan 24 2006 - 13:14:13 GMT