Automatic Semantic Activation and Episodic Memory Encoding

An experiment was conducted to investigate the influence of a briefly presented patternmasked stimulus on (1) subjects' latency'to make a lexical decision regarding a subsequent letter string and (2) their episodic encoding of that letter string in long-term memory. During the first half of the experiment subjects participated in a primed lexical decision task (LDT). Half of the subjects received the primes at a preexperimentally determined critical threshold, whereas the remaining half received the primes at a suprathreshold level. The primes were either related (GRAPE), neutral (XXXXX), or unrelated (BOX) to the targets (JAM). The results of this priming task indicated that subjects responded faster to words which followed a related prime than to words which followed an unrelated prime in both the suprathreshold and, more interestingly, in the threshold condition, where the subjects were actually unable to reliably report the presence of the prime. Subjects were then given an episodic recognition test for the target words which were presented during the LDT. In the crucial conditions, the target was either contextually paired with the same related priming word or a different related word. For example, if the subject received the prime GRAPE followed by the target JAM, then at recognition the subject either received the pair GRAPE JAM or TRAFFIC JAM, with the task being to simply recognize the second word in each pair. The recognition results indicated that for the suprathreshold condition there was a large deleterious effect of switching context between study and test, whereas, for the threshold condition, there was virtually no effect of switching context. These results were viewed as indicating that it is possible to produce activation in semantic memory via a threshold stimulus, as indicated by the obtained priming effect; however, this activation does not appear to be useful in directing conscious attention for long-term storage.