The present study was designed to test the neodissocation hypothesis of hypnotic amnesia. The predictions made by this hypothesis are based on theoretical distinctions between episodic and semantic memory. Hypnotic amnesia is viewed as an involuntary occurance that renders episodic material inaccessible, and thereby leads to recall deficits on episodic memory tasks. However, the semantic properties of amnesic material supposedly remain unaffected by an amnesia suggestion. For this reason, the neodissociation hypothesis predicts no difference between the responses of amnesic subjects and the responses of non-amnesic subjects on semantic memory tasks.
The present study tested predictions of the neodissociation hypothesis using a paradigm that was based on research results from the area of personality assessment. Research in this area has consistently demonstrated high correlations between certain sets of trait definitions. These findings suggest a highly stable associative network between related traits in semantic memory.
Subjects high in hypnotic susceptibility were randomly assigned to one of five experimental groups. All subjects were hypnotized and then presented with a personality profile defining a hypothetical individual (target) on a set of trait dimensions. For all subjects the target was described with two non-critical traits: "basically undependable" and "equally goodnatured and irritable". Two critical traits ("sociable" and "calm") were experimentally manipulated: a) Subjects in the first group learned all four traits (the two non-critical traits and the two critical traits), and then received an amnesia suggestion for one of the critical traits (sociable). b) Those in the second group also learned all four traits and then received and amnesia suggestion for the second critical trait (calm). c) Subjects in the third group served as a control for the first group; they initially learned the two non-critical traits and the critical trait "sociable". The critical trait "calm" was never presented to this group. No amnesia suggestion was administered to these subjects. d) The fourth group served as a control for the second group. Subjects in this group initially learned both non-critical traits and the critical trait "calm". The critical trait, "sociable", was not presented to this group. No amnesia suggestion was administered, e) The fifth group learned all four traits (the two non-critical and the two critical), but was not administered an amnesia suggestion. This group served as a control for both amnesia groups.
In order to compare the responses of amnesic subjects and subjects not given an amnesia suggestion, subjects were given a booklet containing 12 new bipolar dimensions. Three of these trait dimensions were related to sociable and three trait dimensions were related to calm. Subjects' ratings were compared across groups. These ratings constituted the semantic memory task for the experiment. The neodissociation hypothesis predicted that the critical trait (e.g. sociable) initially learned by subjects would continue to influence their ratings on related trait dimensions (e.g. talkative vs. silent). Therefore, on traits related to sociable, the responses of subjects made amnesic for sociable and the responses of subjects given the critical trait sociable but no amnesia suggestion, should not differ.
Results were inconsistent with this prediction. The responses of amnesic subjects were significantly different from the responses of subjects given the critical trait but no amnesia suggestion. In addition, during the suggestion period, the responses of subjects amnesic for a critical trait did not differ significantly from the responses of subjects never given the critical trait. These results suggest that giving subjects an amnesia suggestion for a critical trait significantly influenced their ratings on related trait dimensions (the semantic memory task). Subjects rated the target as though they had never learned the critical information (i.e. ratings were equivalent to the no critical trait groups). Alternative explanations for these results are discussed.