"6209","Response probability and latency: a straight line, an operational definition of meaning and the structure of short term memory","The functional relationship between response probability and time is investigated in data from Rubin, Hinton and Wenzel (1999) and Anderson (1981). Recall/recognition probabilities and search times are linearly related through stimulus presentation lags from 6 seconds to 600 seconds in the former experiment and for repeated learning of words in the latter. The slope of the response time vs. probability function is related to the meaningfulness of the items used. The Rubin et al data suggest that only one memory structure is present or that all memory structures probed show the same linear relation of response probability and time. Both sets of data also suggest that the memory items, presumably in the neocortex, have a finite effective size that shrinks in a logarithmic fashion as the time since stimulus presentation increases or the overlearning decreases, away from the start of the search. According to the logarithmic decay, the size of the memory items decreases to a couple of neurons at about 1500 seconds for recall and 1100 seconds for recognition – this could be the time scale for a short term memory being converted to a long term memory. The incorrect recall time saturates in the Rubin et al data (it is not linear throughout the experiments), suggesting a limited size of the short term memory structure: the time to search through the structure for recall is 1.7 seconds. For recognition the corresponding time is about 0.4 seconds, to compare with the 0.243 seconds given by the data analysis of Cavanagh of Sternberg-like experiments (1972). ","http://cogprints.org/6209/","Tarnow, Dr. Eugen","UNSPECIFIED"," Tarnow, Dr. Eugen (2008) Response probability and latency: a straight line, an operational definition of meaning and the structure of short term memory. [Journal (On-line/Unpaginated)] ","","2008-09-15"