Neural Firing Patterns Are More Schematic and Less Sensitive to Changes in Background Visual Scenes in the Subiculum than in the Hippocampus

Abstract

Literature suggests that the hippocampus is central to processing visual scenes to remember contextual information, but the roles of its downstream structure, subiculum, remain unknown. Here, single units were recorded simultaneously in the dorsal CA1 and subiculum while male rats made spatial choices using visual scenes as cues in a T-maze. The firing fields of subicular neurons were schematically organized following the task structure, largely divided into pre-choice and post-choice epochs, whereas those of CA1 cells were more punctate and bound to specific locations. When the rats were tested with highly familiar scenes, neurons in the CA1 and subiculum were indistinguishable in coding the task-related information (e.g., scene, choice) through rate remapping. However, when the familiar scenes were blurred parametrically, the neurons in the CA1 responded sensitively to the novelty in task demand and changed its representations parametrically following the physical changes of the stimuli, whereas these functional characteristics were absent in the subiculum. These results suggest that the unique function of the hippocampus is to acquire contextual representations in association with discrete positions in space, especially when facing new and ambiguous scenes, whereas the subiculum may translate the position-bound visual contextual information of the hippocampus into schematic codes once learning is established.SIGNIFICANCE STATEMENT Although the potential functional significance has been recognized for decades for the subiculum, its exact roles in a goal-directed memory task still remain elusive. In the current study, we present experimental evidence that may indicate that the neural population in the subiculum could translate the location-bound spatial representations of the hippocampus into more schematic representations of task demands. Our findings also imply that the visual scene-based codes conveyed by the hippocampus and subiculum may be identical in a well learned task, whereas the hippocampus may be more specialized in representing altered visual scenes than the subiculum.

Keywords: episodic memory; hippocampus; rate remapping; single unit recording; subiculum; visual scene.

Figure 1.

Behavioral paradigms and performance. A, The standard version (STD) of visual scene memory (VSM) task. Scenes (Z, zebra stripes; B, bamboo; P, pebbles; M, mountains) used as visual contexts. The rat chose an arm associated with the visual scene to obtain a piece of cereal in a food well. RW, reward; S, start box. B, Performance for each scene stimulus (color-coded for individual rats). Performance exceed criterion (dashed line, 75%) for all scenes. Box plot indicates range and median value. C, Scene stimuli used in the blurred version of the VSM task and the correct responses associated with the stimuli. Only Z and P were used in the blurred version. The original stimuli (STD or No-Blur) were blurred by applying a Gaussian smoothing either by 30% (Lo-Blur) or 50% (Hi-Blur). D, Behavioral performance in the blurred version of the VSM task. Performance (black line) significantly dropped only in the Hi-Blur condition compared with all other conditions. However, the pixel-to-pixel correlation coefficient between original and blurred scenes decreased almost linearly as the amount of blur increased (red line for Z and blue line for P). Mean ± SEM. **p < 0.01.

Figure 2.

Simultaneous recording of single units from the CA1 and subiculum. A, Proportional distribution of cells recorded in the CA1 (blue) and subiculum (SUB; red) along the proximo-distal axis (top) and a flat map showing tetrode positions in the CA1 and subiculum (bottom). The intermediate transition zone (SUB/CA1) is shown in white. Numbers on the left of the flat map indicate relative positions (mm) from bregma. Colored dots represent tetrode positions for individual rats. A, anterior; P, posterior; M, medial; L, lateral. B, Nissl-stained photomicrographs of the tissue sections that contained the tetrode trajectories marked by the arrows in A. C, Basic firing properties of single units in the CA1 and subiculum.

Figure 3.

Event-related firing patterns in the CA1 and subiculum. A, Major events in the VSM task. The opening of the start box door (S), turning to the left or right arm at the choice point (C), and reaching the reward location (R) were defined as three major events. Three sensor-crossing points in the stem and bisecting point of each arm were minor events. The arrow denotes the running direction. B, Event rate map (ERM). The raster plot of a single unit is used to as an example to illustrate how spiking activities were grouped into discrete event epochs to result in the ERM. C, Examples of the ERMs and spatial rate maps (SRM) of cells in the CA1 and subiculum. The epoch between the opening of the start box and track entrance was not represented in the spatial rate map because the associated position data were not recorded in the start box in our experimental settings, whereas the event rate map could represent neural activity in the entire task period (including the neural activity inside the start box). Although the information-organizing schemes were different between the event rate map (time) and spatial rate map (location), both formats were very similar because time and location were highly correlated in our task on the maze. D, Distribution of the Pearson correlation coefficients between the event rate maps and spatial rate maps (excluding the start box) of individual neurons of the CA1 (left) and subiculum (right). Two type of maps were highly correlated in the most of cells, and only 6 of 325 cells (n = 1/109 in CA1, n = 5/216 in subiculum) showed no correlation between two maps.

Figure 4.

Types of firing fields in the CA1 and subiculum. A, B, Representative ERMs in the CA1 (A) and subiculum (B). Numbers indicate the maximal firing rates. The color bar denotes the color scale for firing rate (max, Maximal firing rate). C, The proportion of SF and MF types in the CA1 and subiculum. D, Comparison of the field sizes of the units with SFs and MFs in the CA1 and subiculum. Inset, Same data presented as bar graphs. Median ± 95% confidence interval/2. **p < 0.01, ***p < 0.001. E, Histograms to compare the proximo-distal distributions of SF units with MF units in the subiculum. Overlapping areas between SF and MF distributions are shown in white. Note the presence of more MF units in the distal subiculum (green areas) and more SF units in the proximal subiculum (red areas).

Figure 5.

Schematic firing patterns of the neural populations in the subiculum, but not in the CA1. A, Population ERMs for the CA1 and subiculum. Individual ERMs were aligned according to their maximal firing locations along the event dimension (S-C-R). The white dotted line denotes the boundary between the pre-choice (pre-C: S to C) and post-choice (post-C: C to R) periods. B, Autocorrelation matrix showing the cross-correlations between the same population ERMs to reveal positively correlated (warm colors) and anti-correlated (cool colors) areas in the population ERM. Black dashed line: choice point. C, Task-congruent (TC, orange triangles) and task-incongruent (TI, blue rectangle) areas in the autocorrelation matrix (only shown in the lower half of the matrix to avoid duplicate information). D, E, Comparison of mean correlation coefficients between the CA1 and subiculum in the TC area (D) and TI area (E). ***p < 0.001. F, Representative examples of the population ERMs containing the randomly shifted ERMs for individual cells in the CA1 and subiculum. The locations of ERMs were shuffled and sorted by their peak firing locations. Note that the characteristic schematic firing across the choice point observed in the original population ERM for the subiculum disappeared. G, Averaged autocorrelation matrix using the shifted ERMs. The correlational pattern made of the shifted ERMs in the subiculum became similar to that of the CA1. H, Representative examples of cells showing speed correlation in the CA1 (left) and subiculum (right). The ERMs of the speed-correlated cells were similar to the speed maps in the corresponding session (color maps, top; numbers indicate maximal speed and firing rate), and there was a significant correlation between the in-field firing rates and their associated running speeds in individual trials (scatter plots, bottom).

Figure 6.

Illustration of the differential coding schemes between the hippocampus and subiculum. Location-specific firing patterns of CA1 cells are translated (by chunking) into schematic representations in the subiculum. Overlapping fields in different colors illustrate the scene-dependent rate remapping with each field associated with one of the visual scenes in the task.

Figure 7.

Task-dependent rate remapping in the CA1 and subiculum. A, B, Representative ERMs of CA1 (A) and subicular (B) cells, associated with four scenes (Z, B, P, and M). Cells are subcategorized into nonspecific, choice-specific, and scene-specific units. Scene and associated spatial choice are labeled on the left side of each ERM. The number at the end of the ERM of each cell denotes the maximum firing rate.

Figure 8.

Comparison of rate remapping between the CA1 and subiculum. A, B, Illustration of the procedures for calculating the RDI. Each panel consists of the ERMs associated with different task-relevant information (top) and line graphs that compare the firing rates between different scene conditions along the event axis (bottom). A, Rate difference index for choice (RDI_CHOICE). Scenes associated with the same choice arm were combined to measure the field's firing rate for each side (i.e., FR_LEFT and FR_RIGHT). RDI_CHOICE is the difference between the firing rates associated with the left and right choices. B, Rate difference index for the scene (RDI_SCN). RDI_SCN-L and RDI_SCN-R are RDI values computed for different scenes that share the common choice arm, and the larger of the two was taken as RDI_SCN of the unit. C, D, Cumulative distributions of RDI_CHOICE (C) and RDI_SCN (D). RDI_CHOICE was not significantly different between the CA1 and subiculum, and the same was the case for RDI_SCN.

Figure 9.

Dramatic neuronal remapping in the CA1 upon task switching, but not in the subiculum. A, B, Comparison of spiking activities between STD and No-Blur conditions in the CA1 (A) and subiculum (B). Spiking activity was turned on (cell 1) or off (cell 2) as task demand shifted from processing original scenes to processing various scenes including the blurred ones from the 21st trial (red dashed line). Only the spiking data from the STD and No-Blur trials are shown here. Such radical remapping was less frequently observed in the subiculum. Neural spikes are aligned to the start-box door-opening event (0). The ERMs for STD and No-Blur blocks are shown. For each cell, the average waveforms from four channels of the tetrode recorded from pre-sleep to post-sleep session are shown. Scale bar, 50 μV. C, Comparison of the amounts of rate modulation between STD and No-Blur conditions in the CA1 and subiculum. ***p < 0.001. D, The proportion of cells that underwent significant rate modulations between STD and No-Blur. The numbers of cells are given in the parentheses. *p < 0.05.

Figure 10.

Blur level-dependent neural changes in the CA1, but not in the subiculum. A, B, Representative ERMs of CA1 (A) and subiculum (B). The amount of rate modulation (RDI_SCN-C) between different scene/choice decreased in CA1 units as the level of noise (i.e., blur) in the scene stimuli increased, whereas the patterns of rate modulation in the subiculum were not correlated with the blur levels of the stimuli. The numbers below the ERM indicate the maximal firing rates. C, Cumulative distributions of RDI_SCN-C for different blur conditions in the CA1 and subiculum. Note the gradual shift toward lower RDI_SCN-C values as the blur level increased in the CA1, but not in the subiculum. Inset, The same data shown as bar graphs. Median ± 95% confidence interval/2. **p < 0.01. D, Comparing RDI_SCN-C values of No-Blur and Lo-Blur conditions between the CA1 and subiculum. More units in the CA1 changed their firing rates as a function of visual scenes than in the subiculum. *p < 0.05.

Figure 11.

VSM task with masked stimuli. A, Scene stimuli used in the masking experimental session. In each session, one of the two pairs was chosen from mask patterns 1 and 2, and those individual masking patterns were pseudorandomly presented in an intermixed fashion throughout the session with the original scene. B, Box plots showing rat's performance in the masking session. For each masking pattern with either larger or small viewing holes, mean performance of a rat is plotted as a dot. Dashed line: chance performance level. Box plot indicates range and median value. **p < 0.01.