Dual Coding with STDP in a Spiking Recurrent Neural Network Model of the Hippocampus

Abstract

The firing rate of single neurons in the mammalian hippocampus has been demonstrated to encode for a range of spatial and non-spatial stimuli. It has also been demonstrated that phase of firing, with respect to the theta oscillation that dominates the hippocampal EEG during stereotype learning behaviour, correlates with an animal's spatial location. These findings have led to the hypothesis that the hippocampus operates using a dual (rate and temporal) coding system. To investigate the phenomenon of dual coding in the hippocampus, we examine a spiking recurrent network model with theta coded neural dynamics and an STDP rule that mediates rate-coded Hebbian learning when pre- and post-synaptic firing is stochastic. We demonstrate that this plasticity rule can generate both symmetric and asymmetric connections between neurons that fire at concurrent or successive theta phase, respectively, and subsequently produce both pattern completion and sequence prediction from partial cues. This unifies previously disparate auto- and hetero-associative network models of hippocampal function and provides them with a firmer basis in modern neurobiology. Furthermore, the encoding and reactivation of activity in mutually exciting Hebbian cell assemblies demonstrated here is believed to represent a fundamental mechanism of cognitive processing in the brain.

Calcium control of triphasic hippocampal STDP

“…Recent computational modelling suggests that the interaction of these processes is critical to establish and maintain appropriate conditions for transient dynamics during cognitive processing, and the examination of a unified model of neural and synaptic plasticity is therefore a critical direction for future theoretical studies [86][87][88][89]. More generally, an examination of the synaptic and neural dynamics generated by the triphasic STDP rule in network models of hippocampal function with more realistic activity patterns, including theta modulation and phase precession, would contribute significantly to the understanding of hippocampal function during putative learning behaviour [90].…”

From Blickets to Synapses: Inferring Temporal Causal Networks by Observation

“…5). This mechanism is based on the well-documented phenomena of theta coding in the hippocampus, which is generated by the phase precession of place cell activity relative to the ongoing 4-8 Hz theta-rhythm (Bush, Philippides, et al, 2010;Huxter, Senior, Allen, & Csicsvari, 2008;Lisman, 2005;O'Keefe & Recce, 1993), although the same mechanism is possible with any underlying oscillation frequency (e.g., gamma phase coding; Fries, Nikolic, et al, 2007). Theta coding allows a temporal relationship between behavioral events (i.e., the traversal of neighboring place fields) that is too long to be captured by the effective time window of synaptic plasticity to be represented in a compressed form and therefore encoded in synaptic weights.…”

“…Conversely, during recall, a decreased level of ACh enhances synaptic transmission, to facilitate the retrieval of previously learned associations, while suppressing synaptic plasticity to prevent interference. There is considerable biological evidence for this effect, and several previous neuronal network models have incorporated the same abstract mechanism (Bush et al, 2010;Hasselmo, 2006;Yu & Dayan, 2002). Here, we implement putative ACh neuromodulation by simply increasing the gain of synaptic weights in the TCN by a factor of 5 during the recall period, such that the causal relationships encoded in the TCN can be sampled.…”

“…It is the temporal counterpart of the spatial redundancy achieved before using a population of neurons to represent each event. How such theta-phase coding is generated in the brain is a matter of ongoing debate, and we do not attempt to account for the explicit mechanism here, but simply assume that such a mechanism exists, in accordance with empirical data and previous modeling studies (Bush et al, 2010;Nadasdy, Hirase, et al, 1999;O'Keefe & Recce, 1993). In our phenomenological account, the temporal sequence of events in a trial is encoded as a spatiotemporal input spike pattern that is repeated at a rate of 8 Hz, with the relative timing of each event encoded in the relative phase offset of the corresponding input spike.…”

Generalization of learning by synchronous waves: from perceptual organization to invariant organization

“…Learning models that include both tonic activation and individual spiking within the formulation of the STDP rule show hetero-associative properties (Bush et al 2010). …”