proportion of correct recalls drops off monotonically with serial position. This is exactly opposite to item recognition where recency predominates (see Fig. 2). Typical serial-position curves for recall are shown in Fig. 10, and the results for auditory and visual presentation are quite similar. Note that, at each serial position, the proportion of correct recalls is inversely proportional to list length.
I have run a number of simulations to see if the chunking model can mimic these data, and it certainly can. Typical results are shown in Fig. 11. With one exception (Serial Position 6) the proportion of correct recalls is again proportional to list length.
For each list length we formed an n-chunk according to Eq. 3. We then unpacked the chunk item by item using the multiple-convolution algorithm. At each step in the process we scored the best match as the item recalled. If all matches were negative we scored on omission. What is retrieved at each step must enter into the retrieval cue for the next step. If the recall was successful we used the target item; otherwise we used whatever item had been recalled. For further details see Murdock ( 1995).
While the serial-position curves from these simulations were quite good, further analyses did reveal a problem. The error gradients either showed primacy or were approximately flat. In experimental data error gradients are usually sharply peaked around the target serial position ( Lee & Estes, 1977; Naime, 1991). One possible reason for this discrepancy is that experimental subjects group the longer lists into two chunks and the intrusions generally come from the appropriate chunks. In the simulations we used a single chunk even for the longest lists.
In this chapter I have presented a general account of TODAM, a theory of distributed associative memory that deals with the storage and retrieval of item, associative, and serial-order information. It attempts to explain behavioral data from psychological studies of short-term episodic memory. It is reasonably successful in explaining many patterns of data, and hopefully can be useful in understanding the role of memory in some of the broader issues that are the theme of this conference.
Anderson, J. A. ( 1970). "Two models for memory organization using interacting traces". Mathematical Biosciences, 8, 137-160.
Anderson, J. A. ( 1973). "A theory for the recognition of items from short memorized lists". Psychological Review, 80, 417-438.
Arbib, M. A. (Ed.). ( 1995). The handbook of brain theory and neural networks. Cambridge, MA: MIT Press.
Barnes, J. M., & Underwood, B. J. ( 1959). "Fate" of first-list associations in transfer theory. Journal of Experimental Psychology, 58, 97-105.
Borsellino, A., & Poggio, T. ( 1973). "Convolution and correlation algebras". Kybernetik, 122, 113-122.
Clarke, S. E., & Grunlund, S. D. ( 1996). "Global matching models of recognition memory: How the models match the data". Psychonomic Bulletin & Review, 3, 37-60.
Questia, a part of Gale, Cengage Learning. www.questia.com
Publication information: Book title: Brain and Values:Is a Biological Science of Values Possible. Contributors: Karl H. Pribram - Editor. Publisher: Lawrence Erlbaum Associates. Place of publication: Mahwah, NJ. Publication year: 1998. Page number: 308.
This material is protected by copyright and, with the exception of fair use, may not be further copied, distributed or transmitted in any form or by any means.