Mathematical Psychology
Psychometrics & MeasurementDavid KrantzPatrick SuppesFrederic Lord
About
HomeKey FiguresLearning & Memory Modelers
Learning & Memory Modelers

Richard Shiffrin

Richard Shiffrin (b. 1942) co-developed the SAM and REM models of memory and, with Atkinson, proposed the influential modal model of memory with distinct short-term and long-term stores.

Richard M. Shiffrin, working at Indiana University for over five decades, has been one of the most productive mathematical modelers of memory and attention. His early collaboration with Richard Atkinson produced the Atkinson-Shiffrin model (the "modal model"), which distinguished sensory, short-term, and long-term memory stores. His later work with Jeroen Raaijmakers produced the SAM model, and his REM model provided a Bayesian account of recognition memory.

The SAM Model

SAM: Search of Associative Memory P(sample item i) = S(Q, i) / Sum_j S(Q, j)

P(recover item i | sampled) = 1 - exp(-S(Q, i))

S(Q, i) = Product_k W_k(Q_k, i) (cue-item strengths)
Q = set of retrieval cues, W = associative strength

SAM (Raaijmakers & Shiffrin, 1981) formalizes retrieval as a two-stage process: probabilistic sampling of items from memory (based on relative cue-item strength) followed by recovery of sampled items (based on absolute strength). This architecture explains list-strength effects, part-list cuing inhibition, the distinction between recall and recognition, and output interference.

The Atkinson-Shiffrin Model

The 1968 Atkinson-Shiffrin model proposed that information flows through three stores: a brief sensory register, a limited-capacity short-term store (where rehearsal maintains information), and a long-term store. Although details have been revised, this architecture remains the conceptual framework within which memory research is organized and taught.

The REM Model

The REM (Retrieving Effectively from Memory) model (Shiffrin & Steyvers, 1997) provided a Bayesian account of recognition memory. Items are stored as noisy feature vectors, and recognition involves computing a likelihood ratio comparing the probability that a test item matches a stored trace versus being a new item. REM naturally explains the mirror effect (high-frequency items produce both fewer hits and fewer false alarms), list-length effects, and the word-frequency effect in recognition.

Legacy and Impact

Shiffrin's models have set the standard for formal theories of memory retrieval. The SAM framework spawned a family of models (including REM and others) that account for an extraordinary range of memory phenomena within a unified mathematical framework. His insistence on quantitative precision and his development of computational tools for fitting complex models to behavioral data have influenced multiple generations of memory researchers.

Related Topics

Memory ModelsSAM ModelMemory ModelsREM ModelMemory ModelsWorking Memory Capacity

References

  1. Raaijmakers, J. G. W., & Shiffrin, R. M. (1981). Search of associative memory. Psychological Review, 88(2), 93-134. doi:10.1037/0033-295X.88.2.93
  2. Atkinson, R. C., & Shiffrin, R. M. (1968). Human memory: A proposed system and its control processes. In K. W. Spence & J. T. Spence (Eds.), The psychology of learning and motivation (Vol. 2, pp. 89-195). Academic Press. doi:10.1016/S0079-7421(08)60422-3
  3. Shiffrin, R. M., & Steyvers, M. (1997). A model for recognition memory: REM -- retrieving effectively from memory. Psychonomic Bulletin & Review, 4(2), 145-166. doi:10.3758/BF03209391
  4. Malmberg, K. J. (2008). Recognition memory: A review of the critical findings and an integrated theory for relating them. Cognitive Psychology, 57(4), 335-384. doi:10.1016/j.cogpsych.2008.02.004

External Links