Global memory model

from Wikipedia, the free encyclopedia

In the 1980s, when the weaknesses of the multi-memory models became more apparent, several so-called global memory models were developed such as:

  • MINERVA 2 (Hintzman, 1984 ),
  • SAM - Search of Associative Memory (Gillund & Shiffrin, 1981),
  • TODAM - Theory of Distributed Associative Memory (Murdock, 1982),
  • CHARM - Composite holographic associative recall model (Metcalfe Eich, 1982) and
  • the matrix model (Pike, 1984).

Although all of these models are very different, there are fundamental similarities that set them apart from the class of multi-memory models and at the same time together fall into the category of global memory models.

The reason for this arises from the basic axioms , more precisely from the assumption of a global “matching” process. According to Marrs (1982) classification approach in cognitive psychology, these are computational models, which primarily allow qualitative predictions using computer simulations . However, it was later shown that, with some modifications, model adjustments with quantifiable fit indices are possible for these models.

The mathematical approach of these models lies in the formalization of their assumptions. Storage processes, retrieval processes and the form of memory representation are formally specified. This is a clear advantage over less formal theories, as predictions and explanations can be derived directly analytically. Thus, more rigorous tests of such formalized theories are possible. This is usually done using computer simulations, the results of which are then compared with data obtained from analog experiments with human test subjects.

The basic common assumption of all these models is that all memory traces present in the memory play a role in both the recognition and the recollection of a particular item. In contrast to local storage models, which mostly assume a search process that compares each memory representation with the test item, a global "matching" or comparison process is postulated instead, which compares the similarity of all items represented in the memory with the test item. This process is parallel and immediate for all memory traces. The result of this global process is a scalar similarity value that can be interpreted as a familiarity value. This familiarity value is used as a basis for many decision-making processes at the memory level, assuming a criteria value with which it is compared. In this way, for example, findings in the field of schema abstraction, frequency assessment and other recognition and recollection paradigms as well as phenomena from thought psychology could be successfully explained.

literature

  • DL Hintzman: MINERVA 2: A simulation model of human memory . In: Behavior Research Methods, Instruments, & Computers , 1984, 16, pp. 96-101.
  • G. Gillund, RM Shiffrin: A retrieval model for both recognition and recall . In: Psychological Review , 1984, 91, pp. 1-65.
  • Janet Metcalfe Eich: A Composite Holographic Associative Recall Model . In: Psychological Review , 1982, 89, pp. 627-661
  • BB Murdock: A theory for the storage and retrieval of item and associative information . Psychological Review , 1982, 89, pp. 609-626.
  • R. Pike: Comparison of convolution and matrix distributed memory systems for associative recall and recognition . In: Psychological Review , 1984, 91 (3), pp. 281-293