Sequences of 6 letters of the alphabet were visually presented for immediate recall to 387 subjects. Errors showed a systematic relationship to original stimuli. This is held to meet a requirement of the decay theory of immediate memory.The same letter vocabulary waa used in a test in which subjects were required to identify the letters spoken against a white noise background. A highly significant correlation waa found between letters which confused in the listening test, and letters which confused in recall.The role of neurological noise i n recall is discussed in relation to these results. It is further argued that information theory is inadequate to explain the memory span, since the nature of the stimulus set, which can be defined quantitatively, as well as the information per item, is likely to be a determining factor.

Immediately after visual presentation, subjects were required to recall 6-letter sequences. Sequences were drawn from four vocabularies. There were two 3-letter vocabularies, distinguished by the probability of acoustic confusion within them, and two 9-letter vocabularies similarly distinguished. Memory span is shown to be effectively independent of information per item, and to depend substantially on the probability of acoustic confusion within vocabularies.A recent paper by the author (Conrad, 1964) demonstrated a highly significant association between errors in immediate recall of 6-letter sequences, and listening errors. Letters of the alphabet which are difficult to hear against a white noise background are also difficult to recall, even when presentation for recall is visual. The effect was also shown to hold true for words (Conrad, 1963).It was argued that, if letters which are acoustically confusing are more likely to be confused in memory, then sequences drawn from acoustically homogeneous vocabularies would be more difficult to recall than those drawn from vocabularies which were acoustically heterogeneous. It was suggested that the acoustic nature of the vocabulary might in fact be a more important variable in determining memory span than the size of vocabulary from which sequences were drawn. The present experiment tests these two hypotheses. METHODThirty-six 6-letter sequences were constructed from each of four consonant 'vocabularies '. The vocabularies were :The sequences were based on tables of random numbers with the following constraints : The vocabularies were chosen on the basis of two variables. First, vocabulary size, three or nine letters. Secondly, acoustic confusability within the vocabulary, so that for each size of vocabulary there was one which had high, and one which had low, confusability. This variable requires explanation.Conrad (1964) presents a confusion matrix for letters of the alphabet spoken against a white noise background. For subsets of this matrix composed of the letters of the

No abstract

No abstract

Results are presented which confirm data already published by Corballis (J 966). and Murray. (J 966J. These show that when a sequence of letters or digits is presented for immediate recall, there is a marked difference in the serial position curve of errors according to whether presentation is auditory or visual. This difference specifically shows as a virtual absence of recency in the visual presentation condition.