Critical role of the hippocampus in memory for sequences of events

Abstract

Recent models of hippocampal function emphasize the potential role of this brain structure in encoding and retrieving sequences of events that compose episodic memories. Here we show that hippocampal lesions produce a severe and selective impairment in the capacity of rats to remember the sequential ordering of a series of odors, despite an intact capacity to recognize odors that recently occurred. These findings support the hypothesis that hippocampal networks mediate associations between sequential events that constitute elements of an episodic memory.

Fig. 1

Sequential order and recognition tasks. Top, sequence of events in each trial. Bottom, an example trial for the sequential order and recognition probes. A–E designates the order of presentation for odors in each series, with odor A presented first and odor E presented last (see Methods). +, reinforced odor; −, nonreinforced odor.

Fig 2

Performance (mean percent correct ± s.e.m.) of control rats and rats with hippocampal lesions on sequential order memory task. Rats with hippocampal lesions were reliably impaired (repeated-measures ANOVA, F_1,12 = 39.609, P < 0.0001) and the severity of impairment did not vary reliably among the probe types (Group × Probe interaction, F_5,60 = 0.472, P = 0.795). Post-hoc tests revealed significant impairment on every type of probe (all t-tests > critical t₁₂ = 2.119; *, P < 0.05). Controls performed significantly above chance on every probe (all one-sample t-tests > critical t₆ = 2.447), whereas rats with hippocampal lesions performed better than chance only on the probe that involved the first versus the last odor in the sequence (t₆ = 6.000; for all other probes, t < 2.447). Performances on different probes are grouped according to the lag (number of intervening elements) between items in the probe test.

Fig. 3

Performance (mean percent correct ± s.e.m.) of control rats and rats with hippocampal lesions on recognition for each of the different types of probes. ‘X’ designates a randomly selected odor that was not presented in the series and used as the alternative choice. Control rats and rats with hippocampal lesions performed equally well on odor recognition (repeated-measures ANOVA, F_1,12 = 0.328, P = 0.577). In addition, recognition scores were higher for more recently presented items (main effect of probe type, F_4,48 = 4.778, P = 0.003), and there was no significant difference between groups in the effect of item recency (Group × Probe interaction, F_4,48 = 0.664, P = 0.620).

Fig. 4

Comparison of performance on recognition and sequential order tests. Control and lesioned rats performed differently between the two tasks (Group × Problem interaction, F_1,12 = 7.385, P = 0.019). Post-hoc tests confirmed that controls did not perform differently on these probes (t₆ = 0.241, P = 0.818). In contrast, rats with hippocampal lesions were impaired on the sequential order probes (t₁₂ = 3.125; *,P = 0.009) but not on the recognition probes (t₁₂ = 0.661, P = 0.521).

Fig. 5

Hippocampal subdivisions affected by the lesions at three anterior–posterior levels. Left, anatomically labeled coronal sections adapted from Swanson. Right, reconstructions of the smallest (dark shading) and largest (light shading) hippocampal lesions. CA1 and CA3, components of Ammon’s horn; DG, dentate gyrus; dSub and vSub, dorsal and ventral components of the subiculum; ENTmv, medial–ventral entorhinal cortex).