The connections of neocortical pyramidal cells can implement the learning of new categories, attractor memory, and top–down recall and attention

Neocortical pyramidal cells have three key classes of excitatory input: forward inputs from the previous cortical area (or thalamus); recurrent collateral synapses from nearby pyramidal cells; and backprojection inputs from the following cortical area. The neocortex performs three major types of computation: (1) unsupervised learning of new categories, by allocating neurons to respond to combinations of inputs from the preceding cortical stage, which can be performed using competitive learning; (2) short-term memory, which can be performed by an attractor network using the recurrent collaterals; and (3) recall of what has been learned by top–down backprojections from the following cortical area. There is only one type of excitatory neuron involved, pyramidal cells, with these three types of input. It is proposed, and tested by simulations of a neuronal network model, that pyramidal cells can implement all three types of learning simultaneously, and can subsequently usefully categorise the forward inputs; keep them active in short-term memory; and later recall the representations using the backprojection input. This provides a new approach to understanding how one type of excitatory neuron in the neocortex can implement these three major types of computation, and provides a conceptual advance in understanding how the cerebral neocortex may work.

The hippocampus, memory, and spatial function

The hippocampal system provides a beautiful example of how different classes of neuronal network in the brain work together as a system to implement episodic memory, the memory for particular recent events. The hippocampus contains spatial view neurons in primates including humans, which provide a representation of locations in viewed space. These representations can be combined with object and temporal representations to provide an episodic memory about what happened where and when. A key part of the system is the CA3 system with its recurrent collateral connections that provide a single attractor network for these associations to be learned. The computational generation of time, encoded by time cells in the hippocampus, is described, and this leads to a theory of hippocampal replay and reverse replay. The computational operation of a key part of the architecture, the recall of memories to the neocortex, is described.

Recurrent cortical circuits implement concentration-invariant odor coding

Animals rely on olfaction to find food, attract mates, and avoid predators. To support these behaviors, they must be able to identify odors across different odorant concentrations. The neural circuit operations that implement this concentration invariance remain unclear. We found that despite concentration-dependence in the olfactory bulb (OB), representations of odor identity were preserved downstream, in the piriform cortex (PCx). The OB cells responding earliest after inhalation drove robust responses in sparse subsets of PCx neurons. Recurrent collateral connections broadcast their activation across the PCx, recruiting global feedback inhibition that rapidly truncated and suppressed cortical activity for the remainder of the sniff, discounting the impact of slower, concentration-dependent OB inputs. Eliminating recurrent collateral output amplified PCx odor responses rendered the cortex steeply concentration-dependent and abolished concentration-invariant identity decoding.

Recurrent cortical circuits implement concentration-invariant odor coding

Animals rely on olfaction to find food, attract mates and avoid predators. To support these behaviors, animals must reliably identify odors across different odorant concentrations. The neural circuit operations that implement this concentration invariance remain unclear. Here we demonstrate that, despite concentration-dependence in olfactory bulb (OB), representations of odor identity are preserved downstream, in piriform cortex (PCx). The OB cells responding earliest after inhalation drive robust responses in a sparse subset of PCx neurons. Recurrent collateral connections broadcast their activation across PCx, recruiting strong, global feedback inhibition that rapidly suppresses cortical activity for the remainder of the sniff, thereby discounting the impact of slower, concentration-dependent OB inputs. Eliminating recurrent collateral output dramatically amplifies PCx odor responses, renders cortex steeply concentration-dependent, and abolishes concentration-invariant identity decoding.

A transformation from temporal to ensemble coding in a model of piriform cortex

Different coding strategies are used to represent odor information at various stages of the mammalian olfactory system. A temporal latency code represents odor identity in olfactory bulb (OB), but this temporal information is discarded in piriform cortex (PCx) where odor identity is instead encoded through ensemble membership. We developed a spiking PCx network model to understand how this transformation is implemented. In the model, the impact of OB inputs activated earliest after inhalation is amplified within PCx by diffuse recurrent collateral excitation, which then recruits strong, sustained feedback inhibition that suppresses the impact of later-responding glomeruli. We model increasing odor concentrations by decreasing glomerulus onset latencies while preserving their activation sequences. This produces a multiplexed cortical odor code in which activated ensembles are robust to concentration changes while concentration information is encoded through population synchrony. Our model demonstrates how PCx circuitry can implement multiplexed ensemble-identity/temporal-concentration odor coding.

A transformation from temporal to ensemble coding in a model of piriform cortex

ABSTRACTDifferent coding strategies are used to represent odor information at various stages of the mammalian olfactory system. A temporal latency code represents odor identity in olfactory bulb (OB), but this temporal information is discarded in piriform cortex (PCx) where odor identity is instead encoded through ensemble membership. We developed a spiking PCx network model to understand how this transformation is implemented. In the model, the impact of OB inputs activated earliest after inhalation is amplified within PCx by diffuse recurrent collateral excitation, which then recruits strong, sustained feedback inhibition that suppresses the impact of later-responding glomeruli. Simultaneous OB-PCx recordings indicate that indeed, over a single sniff, the earliest-active OB inputs are most effective at driving PCx activity. We model increasing odor concentrations by decreasing glomerulus onset latencies while preserving their activation sequences. This produces a multiplexed cortical odor code in which activated ensembles are robust to concentration changes while concentration information is encoded through population synchrony. Our model demonstrates how PCx circuitry can implement multiplexed ensemble-identity/temporal-concentration odor coding.

Desynchronization of Glutamate Release Prolongs Synchronous CA3 Network Activity

Periodic bursts of activity in the disinhibited in vitro hippocampal CA3 network spread through the neural population by the glutamatergic recurrent collateral axons that link CA3 pyramidal cells. It was previously proposed that these bursts of activity are terminated by exhaustion of releasable glutamate at the recurrent collateral synapses so that the next periodic burst of network activity cannot occur until the supply of glutamate has been replenished. As a test of this hypothesis, the rate of glutamate release at CA3 axon terminals was reduced by substitution of extracellular Ca2+ with Sr2+. Reduction of the rate of glutamate release reduces the rate of depletion and should thereby prolong bursts. Here we demonstrate that Sr2+ substitution prolongs spontaneous bursts in the disinhibited adult CA3 hippocampal slices to 37.2 ± 7.6 (SE) times the duration in control conditions. Sr2+ also decreased the probability of burst initiation and the rate of burst onset, consistent with reduced synchrony of glutamate release and a consequent reduced rate of spread of excitation through the slice. These findings support the supply of releasable glutamate as an important determinant of the probability and duration of synchronous CA3 network activity.

Convulsant and Anticonvulsant Effects on Spontaneous CA3 Population Bursts

This paper analyzes the effects of a convulsant and an anticonvulsant manipulation on spontaneous bursts in CA3 pyramidal cells in the in vitro slice preparation under conditions of low (3.3 mM [K+]o) and high (8.5 mM [K+]o) burst probability. When burst probability was low, the anticonvulsant, pentobarbital, produced the anticipated effects: the burst duration decreased and interburst interval increased. However, when burst probability was high, both anticonvulsant and convulsant manipulations decreased the interburst interval and the burst duration. To reconcile these findings, we utilized a model in which CA3 burst duration is limited by activity-dependent depression of CA3 excitatory recurrent collateral synapses and the interburst interval is determined by the time required to recover from this depression. We defined the burst end threshold as the level of synaptic depression at which bursts terminate, and the burst start threshold as the level of synaptic depression at which burst initiation is possible. Synapses were considered to oscillate between these thresholds. When average burst duration and interburst interval data were fit using this model, the paradoxically similar effects of the convulsant and anticonvulsant manipulations could be quantitatively interpreted. The convulsant maneuver decreased both the burst start and end thresholds. The start threshold decreased more than the end threshold, so that the thresholds were closer together. This decreased the time needed to transition from one threshold to the other, i.e., the interburst interval and burst duration. The anticonvulsant manipulation primarily increased the burst end threshold. This also decreased the difference between thresholds, decreasing both interburst interval and burst duration. This model resolves the paradoxical proconvulsant effects of pentobarbital in the CA3 preparation and provides insights into the effects of anticonvulsants on epileptiform discharges in the human EEG.