Neural Population Evidence of Functional Heterogeneity along the CA3 Transverse Axis: Pattern Completion versus Pattern Separation

Abstract

Summary Classical theories of associative memory model CA3 as a homogeneous attractor network because of its strong recurrent circuitry. However, anatomical gradients suggest a functional diversity along the CA3 transverse axis. We examined the neural population coherence along this axis, when the local and global spatial reference frames were put in conflict with each other. Proximal CA3 (near the dentate gyrus), where the recurrent collaterals are the weakest, showed degraded representations, similar to the pattern separation shown by the dentate gyrus. Distal CA3 (near CA2), where the recurrent collaterals are the strongest, maintained coherent representations in the conflict situation, resembling the classic attractor network system. CA2 also maintained coherent representations. This dissociation between proximal and distal CA3 provides strong evidence that the recurrent collateral system underlies the associative network functions of CA3, with a separate role of proximal CA3 in pattern separation.

Topography of Place Maps along the CA3-to-CA2 Axis of the Hippocampus

“…By showing that neural activity patterns in proximal part of CA3 are more discrete than those at more distal locations of the same subfield, the data suggest that pattern separation is more likely to take place at the proximal end, and pattern completion more likely at the distal end. The observations are consistent with claims from a lesion study (Hunsaker et al, 2008), an immediate-early gene activation study (Marrone et al, 2014), and a neural ensemble recording study published back-to-back with the present paper (Lee et al, 2015), which all report stronger pattern-separation properties at the proximal end of CA3. The present work shows that the generalization gradient is implemented as an enhancement in the overlap of firing patterns of place cells at the distal end of CA3, and it further identifies a continuation of this functional gradient into the adjacent CA2 region.…”

What Learning Systems do Intelligent Agents Need? Complementary Learning Systems Theory Updated

“…This process helps minimize the overlap of CA3 patterns for different memories, increasing storage capacity and minimizing interference between them, even if the two memories represent similar events that have highly overlapping patterns in neocortex and ERC. Empirical evidence provides support for this, with one study [137] showing that the representation supported by DG was highly sensitive to small changes in the environment, despite evidence that incoming inputs from the ERC were little affected (also see [133,145]). Furthermore, DG lesions impair an animals' ability to learn to respond differently in two very similar environments while leaving the ability to learn to respond differently in two environments that are not similar [136].…”

“…However, when the environment produces an input on the ERC that has low overlap with patterns stored previously, the DG recruits a new, statistically independent cell population in CA3 (i.e., pattern separation [27]). Emerging evidence suggests that the amount of overlap required for pattern completion (as well as other characteristics of hippocampal processing) may differ across the proximal-distal [145,146] and dorso-ventral axes [98,[147][148][149][150] of the hippocampus, and may be shaped by neuromodulatory factors (e.g., Acetylcholine) [85,151]. Also, incomplete patterns require less overlap with a stored pattern than distorted ones for completion to occur, so that partial cues will tend to produce completion, as when one sees the watering hole and remembers seeing a lion there previously [27].…”

Investigation of the hippocampal information processing in freely moving rats

“…For CA3 neurons, a gradient in spatial selectivity was observed from CA3c (highly selective and informative) toward CA3a, consistent with recent observations (Lee et al, 2015. Again, this finding is in line with the gradient of changing connectivity in this region.…”

“…CA3c has stronger connections with the CA1 region and is relatively more influenced by perforant path information (which is first processed in the dentate gyrus), while CA3a has the strongest recurrent connectivity and receive the entorhinal information from layer II entorhinal axons both directly via monosynaptic connections, and disynaptically through the dentate gyrus . Analogous to CA1, spatial coding properties have been found to gradually change along the CA3 transverse axis (Lee et al, 2015;.…”