Fig. 63 A taxonomy of long-term memory, adapted from Milner et al. (1998). Selected brain areas that subserve each type of memory system are indicated at the bottom of the diagram.

memory, i.e. "habituation, and "sensitization, subserved by specific reflex pathways; (b) "classical conditioning, which is fUrther classified into conditioning of musculoskeletal reflexes, subserved among others by the "cerebellum, and conditioning of emotional responses, subserved among others by the "amygdala ("fear conditioning); (c) procedural memory, i.e. memory for "habits and "skills, subserved by a corticostriatal system; and (d) "priming, which, in its multiple variants, is believed to be subserved by cortex. It can be readily seen that, whereas the classification of declarative memory is based on a unified concept and seems 'natural', non-declarative memory is defined by exclusion, and its subtaxa are rather heterogeneous.

The Zeitgeist taxonomy of long-term memory systems evokes several types of interrelated problems: 1. Criteria. The notorious 'problem of the criterion' pops up again. In order to classify correctly we must know a lot about the system, but in order to know a lot about the system we must be able to classify it correctly. What shall we use as the decisive criterion for the delineation of a memory system? Common behavioural (phenomenological) properties? The faculties affected by circumscribed brain lesions? The evolutionary history of parts of the brain? As mentioned above, Tulving (1985a) proposes to consider both functional neuroanatomy and behaviour, but the same brain structure could contribute to different tasks, and a seemingly similar behavioural output could reflect different computational goals and "algorithms. For example, perceptual and conceptual priming both involve nonconscious modulation of the processing, retrieval, or production of a mental item by prior exposure to specific information on that item or on items associated with it. But one could come up with very different systems that share this property. So is there a natural 'priming system' in the brain?2

2. Which leads to the issues of *system vs. process and system vs. attribute. Is priming a memory system, or is it a cross-the-board process, or a manifestation of automaticity ofdifferent processes, shared by "skill systems (e.g. Logan 1990)?

3. Which leads to the issue of superfluity. This taps into the tension between the mind's tendency to classify and the continuity of nature (Aristotle, Parts of Animals 642b5-644a12; for selected examples see "amygdala and "limbic system). For it is rather easy and even appealing to come up with new types of memory systems, and the minute they are proposed, to come up with neurobiological and behavioural justification, turning the tentative proposal into a premature fait accompli.

4. Which leads to the other facet of the coin, *general-ization. We could err in grouping into a unified memory system the capabilities of species far remote on the phylogenetic scale. For example, to regard some memory performances in rodents as declarative as opposed to nondeclarative may over-generalize and disregard the accomplishments of the evolution of the primates. 'Implicit' vs. 'explicit' may better fit the tests used, but again, the issue of whether even these terms apply to lower invertebrates, which are on a different branch of evolution, is questionable. Disregarding this difficulty could culminate in the pursuit of "red herrings. Hence taxonomy should be better used a heuristic aid to research rather than a guiding criterion.

Selected associations: Criterion, Dimension, Model,

Memory, System

1On additional aspects of top-down vs. bottom-up processing, see "attention, "binding, and "insight.

2This question makes sense only if we consider definitions 1 and 2 of "system. Definition 3 of "system leaves us with the convenient freedom to declare anything a 'system'.

0 0

Post a comment