The human hippocampus beyond the cognitive map: evidence from a densely amnesic patient

Abstract

We tested a densely amnesic patient (P9), with bilateral hippocampal damage resulting from an autoimmune disorder, and 12 age- and sex-matched controls on a series of memory tasks designed to characterize allocentric spatial learning and memory abilities. We compared P9's ability to perform spatial memory tasks with her ability to perform non-spatial, color memory tasks. First, P9's performance was impaired as compared to controls even in the simplest versions of an allocentric spatial memory task, in which she had to find repeatedly over 10 trials the same location(s) of one, two or three illuminating foot pad(s) among 23 pads distributed in an open-field arena. In contrast, she performed as well as controls when she had to find repeatedly over 10 trials the same one, two or three pad(s) marked by color cue(s), whose locations varied between trials. Second, P9's performance was severely impaired in working memory tasks, when she had to learn on a trial-unique basis and remember the location(s) or the color(s) of one, two or three pad(s), while performing an interfering task during the 1-min interval separating encoding and retrieval. Without interference during the retention interval of the trial-unique tasks, P9's performance was partially preserved in the color tasks, whereas it remained severely impaired in the allocentric spatial tasks. Detailed behavioral analyses indicate that P9's memory representations are more limited than those of controls both in their precision (metric coding) and in the number of items that can be maintained in memory (capacity). These findings are consistent with the theory that the hippocampus contributes to the integration or binding of multiple items, in order to produce high-resolution/high-capacity representations of spatial and non-spatial information in the service of short-term/working and long-term memory.

Keywords: amnesia; declarative memory; interference; medial temporal lobe; memory capacity; parahippocampal; spatial memory.

Figure 1

T2-weighted MRI images of P9's brain showing the extent of the hypersignal reflecting bilateral hippocampal abnormality. (A) MRI performed upon admittance to the hospital following a generalized tonic-clonic epileptic seizure revealed signs of edema over the right hippocampus. (B) MRI performed 2 months following the initial epileptic episode revealed signs of edema over the left hippocampus, and a shrunken right hippocampus. Reproduced from Petel et al. (2011), with permission. (C–F) T2-weighted MRI images of P9's brain performed 2.5 months after the initial episode and showing a shrunken right hippocampus in three different planes: coronal (C), parallel to the long axis of the hippocampus (D) and sagittal (E), and a hypersignal restricted to the left hippocampus in three different planes: coronal (C), parallel to the long axis of the hippocampus (D) and sagittal (F).

Figure 2

T1-weighted MRI images of P9's brain performed 3 years after the onset of pathology showing bilateral hippocampal atrophy, and no obvious signs of pathology in the surrounding cortical areas, from rostral (A) to caudal (D).

Figure 3

Testing environment. (A) Schematic, aerial view of the experimental room (6 × 7 m) containing polarizing features such as doors, windows, tables (white rectangle), chairs, wall posters, etc. Plastic curtains (solid lines) defined three of the boundary walls designating the arena. At each of the four near and far corners of the side walls was a 50 cm gap that served as one of the four different entry points (arrows) through which the subjects must pass in order to enter and exit the testing arena. Twenty-three foot pads were separated by 80 cm from each other and regularly arranged in the arena. (B) Picture of the arena with a participant touching an illuminating foot pad. (C) Picture of the arena with three cued goal locations during the encoding phase in the repeated-trials location condition. (D) Picture of the arena in the color task condition. (E) Picture of the arena with the pole used in the find-and-replace experiment. See main text for detailed description of the experimental room and procedures.

Figure 4

Number of correct choices before the first error (CBE): Number of correct choices before making an error. (A,B) Number of correct choices, absolute average numbers per trial in the repeated-trials and trial-unique conditions. (C,D) Normalized number of correct choices per trial in the repeated-trials and trial-unique conditions: CBE was divided by the number of goal locations. P9's performance was impaired as compared to controls in all conditions, except for the color, repeated-trials condition with one and two goals, and in the trial-unique color condition with one goal and no interference.

Figure 5

Number of errorless trials (NET): number of trials in which subjects made no errors. (A,B) Number of trials. (C,D) Number of trials corrected for the probability to find all goal locations by chance (4.3478% with one location (1/23), 0.3953% with 2 locations (2/23 ^* 1/22), 0.0565% with 3 locations (3/23 ^* 2/22 ^* 1/21). P9's performance was impaired as compared to controls in all conditions, except for the color, repeated-trials condition with one, two and three goals, and in the trial-unique color condition with one goal and no interference.

Figure 6

Number of trials with the first choice correct (FCC). (A,B) Number of trials. (C,D) Number of trials corrected for the probability to find the first goal location by chance (1/23 for one location, 2/23 for two locations, 3/23 for three locations). P9's performance was impaired as compared to controls in all conditions, except for the color, repeated-trials conditions with one, two and three goals, and in the trial-unique color conditions with one, two and three goals and no interference.

Figure 7

Analysis of errors: spatial versus temporal resolution. Types of locations visited when making an error on the first choice upon entering the arena. (A) Analysis for the tasks with one goal location. (B) Analysis for the tasks with two goal locations. (C) Analysis for the tasks with three goal locations. Note that the number of choices of different locations is normalized based on the probability to make that choice.

Figure 8

Find and Replace: the same location was used for 10 repeated trials. P9's performance was impaired as compared to controls. (A) Average distance to the actual goal location. P9 > Controls, t₍₁₁₎ = 9.776, p < 0.0001. (B) Average, minimum and maximal values for subject P9 and 12 age-matched controls. Note that data from the first trial are not included in this analysis. (C) Individual subjects' performance per trial. (D) Average distance to the goal, as a function of the starting location in the arena in the replace phase. The distance to the goal (spatial error) was greater when P9 entered the arena on the same side where the goal was located, as compared to when she entered from the opposite side of the arena. Performance of control subjects did not vary based on the position of the entrance relative to the goal.

Figure 9

Find and replace: a different location was used for each of the 12 different trials. P9's performance was impaired as compared to controls. (A) Average distance to the actual goal location. P9 > Controls, t₍₁₁₎ = 2.503, p = 0.0293. Note that for controls, the distance from the goal was greater in the trial-unique condition, as compared to the repeated trial condition (Figure 8), p = 0.0037. P9's performance did not differ between repeated-trials and trial-unique conditions, p = 0.6656. (B) Average, minimum and maximal values for subject P9 and 12 age-matched controls.