This Week in The Journal

Orientation Maps May Depend on Repulsion in the Retina

Jaeson Jang and Se-Bum Paik

In V1 of primates and carnivores, neurons that respond preferentially to edges of a particular orientation are arranged in orderly patterns: as one moves across the cortical surface, orientation preference gradually shifts. How and why these orientation maps develop is unclear, but one hypothesis is that they depend on the organization of ON and OFF ganglion cells in the retina. ON and OFF cells are evenly spaced in lattices, with their dendritic arbors in separate planes. When such lattices are rotated relative to each other, the offset between neighboring cells changes gradually, producing so-called moiré interference patterns. Paik and Ringach (2011 Nat Neurosci 14:919) hypothesized that if V1 neurons receive inputs primarily from retinotopically nearest neighbors, moiré interference between ON and OFF lattices could create orientation maps. Indeed, by changing the relative periodicity and angle of simulated ON and OFF mosaics, they generated orientation maps similar to those found in various species.

Rotation of two lattices relative to each other produces a moiré interference pattern. Repulsion between ON and OFF retinal ganglion cells can create such patterns, which in turn can explain the formation of orientation maps. See Jang and Paik for details.

If the spatial arrangement of ON and OFF cell mosaics underlies orientation map formation, how are the relative periods and rotation of these mosaics controlled to produce orientation maps with consistent spatial periodicity? Jang and Paik suggest that local repulsive interactions between ON and OFF cells are responsible. Using computer simulations, they showed that repulsion between randomly spaced ON or OFF cells induced positional shifts that created long-range hexagonal mosaics for each cell type. Adding weak repulsive interactions between ON and OFF cells caused the mosaics to rotate relative to each other. Importantly, this weak repulsion was essential for keeping the periodicity of moiré interference patterns in an appropriate range to create realistic orientation maps. Estimations of repulsive energy between pairs of ON and OFF cells in data from experimental animals supported the existence of heterotypic repulsion in vivo.

Together, the data show that repulsive interactions between ON and OFF cells can generate a retinal structure that produces periodic moiré interference patterns, strengthening the plausibility of this mechanism for the establishment of V1 orientation maps in V1. Because heterotypic repulsive interactions are essential for this model to work, future work should verify that such interactions occur and determine whether disrupting them disrupts formation of orientation maps.

Sleep Helps Maintain Pattern-Separation Ability in People

A. Hanert, F.D. Weber, A. Pedersen, J. Born, and T. Bartsch

(see pages 12238–12246)

Each day we encounter many things we have seen before, and others that are new. To function adequately, we must be able to recognize previously seen objects despite contextual variations, but also distinguish new objects from highly similar objects we have encountered before. The hippocampus contributes to these tasks through opposing neural processes called pattern completion (reactivation of previously stored representations when encountering objects with partially overlapping features) and pattern separation (creation of distinct representations for similar objects).

During sleep, reactivation of hippocampal representations is thought to allow transfer of newly encoded information to the cortex, where it is integrated with previous memories to update schemas. These updated schemas are then used for future recognition and discrimination. Hanert et al. asked how this process affects people's pattern-separation ability. To answer this, they had volunteers categorize photographs of common objects, then tested the participants' ability to recognize previously viewed items (targets) and distinguish them from similar but new items (lures), both immediately after the categorization task and 9 h later, with or without sleep in the interim. The proportion of lures that were correctly identified as such was taken as a measure of pattern separation ability; identifying lures as previously seen indicated inadequate pattern separation.

When participants stayed awake after encoding, their pattern separation ability decreased, particularly for lures that were not-so-similar to targets. In contrast, if participants slept after encoding, pattern separation ability was generally maintained at the level measured immediately after encoding. Notably, however, pattern separation for lures that were highly similar to targets was better after wakefulness than after sleep. The authors also found that pattern separation ability after sleep was positively correlated with the density of slow oscillations and spindles recorded electroencephalographically during slow-wave sleep, as well as with phase-locking of theta oscillations to slow oscillations.

Because the photographs used in this study were of common objects, previously stored representations of these objects were likely reactivated during encoding. The results suggest that unique features of these items were initially represented in the hippocampus, but these details degraded during continued wakefulness. Sleep prevented this degradation, thus preserving pattern separation ability. Future work should explore how sleep affects the opposing process, pattern completion.