United We Fall: All-or-None Forgetting of Complex Episodic Events

Do complex event representations fragment over time, or are they instead forgotten in an all-or-none manner? For example, if we met a friend in a café and they gave us a present, do we forget the constituent elements of this event (location, person, and object) independently, or would the whole event be forgotten? Research suggests that item-based memories are forgotten in a fragmented manner. However, we do not know how more complex episodic, event-based memories are forgotten. We assessed both retrieval accuracy and dependency—the statistical association between the retrieval successes of different elements from the same event—for complex events. Across 4 experiments, we show that retrieval dependency is found both immediately after learning and following a 12-hr and 1-week delay. Further, the amount of retrieval dependency after a delay is greater than that predicted by a model of independent forgetting. This dependency was only seen for coherent “closed-loops,” where all pairwise associations between locations, people, and objects were encoded. When “open-loops” were learned, where only 2 out of the 3 possible associations were encoded, no dependency was seen immediately after learning or after a delay. Finally, we also provide evidence for higher retention rates for closed-loops than for open-loops. Therefore, closed-loops do not fragment as a function of forgetting and are retained for longer than are open-loops. Our findings suggest that coherent episodic events are not only retrieved, but also forgotten, in an all-or-none manner.

instances where the element-types acted as a common cue (i.e., Cue-Element) and common retrieval target (i.e., Retrieval-Element). Table S2 provides mean proportion correct (and standard deviations) across modalityspecific elements (i.e., people, locations, and objects) for the common cue and common retrieval target instance at T1 and T2 for Experiment 1 for open-and closed-loops.

Table S2
Mean proportion correct (and standard deviations) across element-types for test sessions T1 and T2 for Experiment 1 for open-and closed-loops. Proportion correct is collapsed across the sleep and awake conditions. For T2, only trials where participants retrieved cue-target associations not previously tested at T1 are included. Element-types (i.e., people, locations, and objects) refer to the type of common cue and retrieval target element (i.e., Direction) The difference in retrieval accuracy across the elements suggests an underlying difference in memory across the elements. However, the difference in retrieval accuracy for the open-loop condition at T2, but not for closedloops, suggests that when dependency is seen (as for the closed-loop condition), accuracy across the elements becomes more symmetrical. Table S3 provides mean proportion correct (and standard deviations) for retrieval accuracy at T2 dependent on whether the pairwise association had been previously tested at T1 or not. A main effect of Sleep was also seen, F(1,100) = 6.54, p = .01, ηp 2 = .06, consistent with the Session x Sleep interaction seen in the main analysis above. No further significant effects or interactions were seen, Fs < 2.58, ps >.11. Prior testing therefore increased performance at T2 relative to associations not previously tested at T1, and this effect was modulated by Sleep and Loop.

Retrieval dependency for tested vs not-tested closed-and open-loops
Mean dependency (and standard deviations) for the data, independent model and dependent model for openand closed-loops, collapsed across sleep, for associations previously tested and not tested are presented in Table   S4 A 2x2x2 (Tested x Loop x Sleep) ANOVA similarly revealed a significant main effect of loop, F(1,100) = 44.29, p < .001, ηp 2 = .31. No other significant effects or interactions were seen, Fs < 1.89, ps > .17.

Dependency across element-and analysis-type
We have previously shown that the proportion of joint retrieval in the observed data does not vary significantly across element-type (i.e., people, locations, and objects), and analysis type (ABAC, where the element A refers to the common cue element-type; and BACA, where the element A refers to common retrieved element-type) (Horner & Burgess, 2014). In order to assess for differences in dependency that might be due to variations in the memorability of element-type across triplets, we repeated the main analysis reported for Experiment 1 across individual element-types for the two separate analysis types (i.e., ABAC and BABC). previously tested vs not tested associations in the awake condition relative to participants in the sleep condition.

Retrieval accuracy for tested vs not-tested closed-and open-loops
A 2x2 (Tested x Loop) ANOVA on retrieval accuracy at T2 revealed a main effect of Tested, F(1,50) = 187.48, p < .001, ηp 2 = .79, with greater accuracy for previously tested than not previously tested associations. No other significant main effects or interactions were observed, Fs < 1.87, ps > .17.