Episodic-like Memory in the Rat

Stephanie J. Babb, Jonathon D. Crystal
2006 Current Biology  
A fundamental question in comparative cognition is whether animals remember unique, personal past experiences [1] [2] [3] . It has long been argued that memories for specific events (referred to as episodic memory) are unique to humans [4, 5] . Recently, considerable evidence has accumulated to show that food-storing birds possess critical behavioral elements of episodic memory [6] [7] [8] [9] [10] , referred to as episodic-like memory in acknowledgment of the fact that behavioral criteria do
more » ... t assess subjective experiences [1] . Here we show that rats have a detailed representation of remembered events and meet behavioral criteria for episodic-like memory. We provided rats with access to locations baited with distinctive (e.g., grape and raspberry) or nondistinctive (regular chow) flavors. Locations with a distinctive flavor replenished after a long but not a short delay, and locations with the nondistinctive flavor never replenished. One distinctive flavor was devalued after encoding its location by prefeeding that flavor (satiation) or by pairing it with lithium chloride (acquired taste aversion [11, 12] ), while the other distinctive flavor was not devalued. The rats selectively decreased revisits to the devalued distinctive flavor but not to the nondevalued distinctive flavor. The present studies demonstrate that rats selectively encode the content of episodic-like memories. Results and Discussion A central feature of human memory is the ability to encode multiple features about previous experiences, in particular, its content (what), location in space (where), and occurrence in time (when) [13] . Scrub jays meet behavioral criteria for episodic-like memory (i.e., flexible deployment of an integrated representation of whatwhere-when an event occurred) [9, 10]. To further facilitate our understanding of the neural mechanisms mediating episodic memory, it would be useful to validate a framework for studying what-where-when memory in rodents. If rats have specific information about the content of events they experienced in the past, together with knowledge of when and where those events occurred, then they should adjust their behavior to the temporal and spatial constraints of food availability. Moreover, they should flexibly change their behavior if a future outcome is expected to be less desirable than in the past [1, 14, 15] . To address this question, we trained rats to discriminate what, where, and when they encountered food [16, 17] . Each day rats were allowed to retrieve food (grape-, raspberry-, or chow-flavored) pellets on an 8-arm radial maze (Figure 1) . After a chow pellet was consumed at an arm, that arm did not provide additional food until the next trial; trials were separated by a day. Rats visited four baited arms (randomly chosen on each trial; study phase; Figure 1A ) and were later returned to the maze after either a short or long retention interval, with all eight locations available (test phase; Figures 1B and 1C ). The first phase provides an opportunity to study the first four locations; the second phase is a test because chow was always available in the test phase at locations that were not visited in the study phase. To add a what component to the task, two studyphase locations were baited with distinctive flavors: one of the locations was baited with grape and the other was baited with raspberry pellets (randomly selected on each study phase), and these locations replenished grape or raspberry pellets, respectively, in the test phase after a long (6 hr) retention interval ( Figure 1C ) but not after a short (1 hr) retention interval (when; Figure 1B) . Chow-flavored pellets were available at previously inaccessible locations in every test phase but never replenished. If rats remembered when and where they had recently encountered distinctive pellets, then they should learn to selectively revisit the distinctive locations after a long delay but withhold revisits after a short delay. Rats were more likely to revisit grape and raspberry locations after long than after short retention intervals ( Figure 2 ; F(1,11) = 120.6, p < .01). Revisit probabilities were similar for both flavors (F(1,11) = 4.63, p > .05), and the effect of the retention interval did not depend on the flavor (F(1,11) = 0.10, p > .05). Accuracy in avoiding revisits to depleted chow-flavored locations was .91 6 .02 and .72 6 .02 (mean 6 SEM) after short and long retention intervals, respectively, consistent with performance declining as a function of retention interval [18] . If the rats remember the specific flavor at each location, then they should flexibly adjust their behavior according to the expected value of foraging at a particular time and place. Therefore, we decreased the value of one distinctive flavor while leaving the value of the other flavors intact. Each rat was satiated to a distinctive flavor during a long retention interval, with the order of flavors counterbalanced across rats. Figure 3 shows the expected probability of revisiting distinctive locations after a long retention interval (baseline) together with the observed probabilities at locations with devalued and nondevalued distinctive flavors; the probability of a revisit differed across conditions (F(2,22) = 7.47, p < .01). Planned comparisons revealed that after a distinctive flavor was devalued, the probability of revisiting the corresponding location on the maze was lower than baseline ( Figure 3; F(1,22) = 14.88, p < .01), but the revisit
doi:10.1016/j.cub.2006.05.025 pmid:16824919 fatcat:us6unow3grgrjoy54vdmb6enbi