Universal memory mechanism for familiarity recognition and identification.

Macaque monkeys were tested on a delayed-match-to-multiple-sample task, with either a limited set of well trained images (in randomized sequence) or with never-before-seen images. They performed much better with novel images. False positives were mostly limited to catch-trial image repetitions from the preceding trial. This result implies extremely effective one-shot learning, resembling Standing's finding that people detect familiarity for 10,000 once-seen pictures (with 80% accuracy) (Standing, 1973). Familiarity memory may differ essentially from identification, which embeds and generates contextual information. When encountering another person, we can say immediately whether his or her face is familiar. However, it may be difficult for us to identify the same person. To accompany the psychophysical findings, we present a generic neural network model reproducing these behaviors, based on the same conservative Hebbian synaptic plasticity that generates delay activity identification memory. Familiarity becomes the first step toward establishing identification. Adding an inter-trial reset mechanism limits false positives for previous-trial images. The model, unlike previous proposals, relates repetition-recognition with enhanced neural activity, as recently observed experimentally in 92% of differential cells in prefrontal cortex, an area directly involved in familiarity recognition. There may be an essential functional difference between enhanced responses to novel versus to familiar images: The maximal signal from temporal cortex is for novel stimuli, facilitating additional sensory processing of newly acquired stimuli. The maximal signal for familiar stimuli arising in prefrontal cortex facilitates the formation of selective delay activity, as well as additional consolidation of the memory of the image in an upstream cortical module.

Optimizing one-shot learning with binary synapses.

A network of excitatory synapses trained with a conservative version of Hebbian learning is used as a model for recognizing the familiarity of thousands of once-seen stimuli from those never seen before. Such networks were initially proposed for modeling memory retrieval (selective delay activity). We show that the same framework allows the incorporation of both familiarity recognition and memory retrieval, and estimate the network's capacity. In the case of binary neurons, we extend the analysis of Amit and Fusi (1994) to obtain capacity limits based on computations of signal-to-noise ratio of the field difference between selective and non-selective neurons of learned signals. We show that with fast learning (potentiation probability approximately 1), the most recently learned patterns can be retrieved in working memory (selective delay activity). A much higher number of once-seen learned patterns elicit a realistic familiarity signal in the presence of an external field. With potentiation probability much less than 1 (slow learning), memory retrieval disappears, whereas familiarity recognition capacity is maintained at a similarly high level. This analysis is corroborated in simulations. For analog neurons, where such analysis is more difficult, we simplify the capacity analysis by studying the excess number of potentiated synapses above the steady-state distribution. In this framework, we derive the optimal constraint between potentiation and depression probabilities that maximizes the capacity.

Impact of spatiotemporally correlated images on the structure of memory.

How does experience modify what we store in long-term memory? Is it an effect of unattended experience or does it require supervision? What role is played by temporal correlations in the input stream? We present a plastic recurrent network in which memory of faces is initially embedded and then, in the absence of supervision, the presentation of temporally correlated faces drastically changes long-term memory. We model and interpret the results of recent experiments and provide predictions for future testing. The stimuli are frames of a morphing film, interpolating between two memorized faces: If the temporal order of presentation of the frame stimuli is random, then the structure of memory is basically unaffected by synaptic plasticity (memory preservation). If the temporal order is sequential, then all image frames are classified as the same memory (memory collapse). The empirical findings are reproduced in the simulated dynamics of the network, in which the evolution of neural activity is conditioned by the associated synaptic plasticity (learning). The results are captured by theoretical analysis, which leads to predictions concerning the critical parameters of the stimuli; a third phase is identified in which memory is erased (forgetting).

Search for fMRI BOLD signals in networks of spiking neurons.

In a recent experiment, functional magnetic resonance imaging blood oxygen level-dependent (fMRI BOLD) signals were compared in different cortical areas (primary-visual and associative), when subjects were required covertly to name images in two protocols: sequences of images, with and without intervening delays. The amplitude of the BOLD signal in protocols with delay was found to be closer to that without delays in associative areas than in primary areas. The present study provides an exploratory proposal for the identification of the neural activity substrate of the BOLD signal in quasi-realistic networks of spiking neurons, in networks sustaining selective delay activity (associative) and in networks responsive to stimuli, but whose unique stationary state is one of spontaneous activity (primary). A variety of observables are 'recorded' in the network simulations, applying the experimental stimulation protocol. The ratios of the candidate BOLD signals, in the two protocols, are compared in networks with and without delay activity. There are several options for recovering the experimental result in the model networks. One common conclusion is that the distinguishing factor is the presence of delay activity. The effect of NMDAr is marginal. The ultimate quantitative agreement with the experiment results depends on a distinction of the baseline signal level from its value in delay-period spontaneous activity. This may be attributable to the subjects' attention. Modifying the baseline results in a quantitative agreement for the ratios, and provided a definite choice of the candidate signals. The proposed framework produces predictions for the BOLD signal in fMRI experiments, upon modification of the protocol presentation rate and the form of the response function.

Mean-field analysis of selective persistent activity in presence of short-term synaptic depression.

Mean-Field theory is extended to recurrent networks of spiking neurons endowed with short-term depression (STD) of synaptic transmission. The extension involves the use of the distribution of interspike intervals of an integrate-and-fire neuron receiving a Gaussian current, with a given mean and variance, in input. This, in turn, is used to obtain an accurate estimate of the resulting postsynaptic current in presence of STD. The stationary states of the network are obtained requiring self-consistency for the currents-those driving the emission processes and those generated by the emitted spikes. The model network stores in the distribution of two-state efficacies of excitatory-to-excitatory synapses, a randomly composed set of external stimuli. The resulting synaptic structure allows the network to exhibit selective persistent activity for each stimulus in the set. Theory predicts the onset of selective persistent, or working memory (WM) activity upon varying the constitutive parameters (e.g. potentiated/depressed long-term efficacy ratio, parameters associated with STD), and provides the average emission rates in the various steady states. Theoretical estimates are in remarkably good agreement with data "recorded" in computer simulations of the microscopic model.

Memory of ordinal number categories in macaque monkeys.

What mechanism underlies serial order memory? Studying preverbal serial memory shows that macaque monkeys reproducing a sequence of items can acquire knowledge of item ordinal position. In our previous experiment, macaques were repeatedly presented with image lists (first shown sequentially and then simultaneously on a touch screen together with a distractor chosen randomly from other lists). The task was to touch list images in the correct order. The monkeys' natural tendency was to categorize images by their ordinal position or number because their most common error was touching the distractor when it had the same ordinal number (in its own list) as the correct image. Item-to-item associations were used to complete the categorization strategy. Proposing a dynamic image-salience hypothesis for serial recall (based on category-to-image influence and a salience computation for identifying touch targets), we now study the category label characteristics in the context of this hypothesis. We found that these category labels are absolute, ordinal-number-based categories (first, second, etc.), not relative memorized as relative distance from the beginning and the end of the list, and not based on fixed ranking of reward contingency/image familiarity. Even isolated from item-item associations, the categories demonstrate category tuning (as well as the corresponding overlap of adjacent ordinal number codes). Moreover, monkeys choose images by proximity of their category to the current touch number, irrespective of the accuracy of the preceding choice. Category tuning itself is symmetric relative to correct ordinal position, but is skewed by other factors (reward, etc.). Tuning width increases with list length, with a concurrent increased use of item-to-item associations for determining touch order.

Learning in realistic networks of spiking neurons and spike-driven plastic synapses.

We have used simulations to study the learning dynamics of an autonomous, biologically realistic recurrent network of spiking neurons connected via plastic synapses, subjected to a stream of stimulus-delay trials, in which one of a set of stimuli is presented followed by a delay. Long-term plasticity, produced by the neural activity experienced during training, structures the network and endows it with active (working) memory, i.e. enhanced, selective delay activity for every stimulus in the training set. Short-term plasticity produces transient synaptic depression. Each stimulus used in training excites a selective subset of neurons in the network, and stimuli can share neurons (overlapping stimuli). Long-term plasticity dynamics are driven by presynaptic spikes and coincident postsynaptic depolarization; stability is ensured by a refresh mechanism. In the absence of stimulation, the acquired synaptic structure persists for a very long time. The dependence of long-term plasticity dynamics on the characteristics of the stimulus response (average emission rates, time course and synchronization), and on the single-cell emission statistics (coefficient of variation) is studied. The study clarifies the specific roles of short-term synaptic depression, NMDA receptors, stimulus representation overlaps, selective stimulation of inhibition, and spike asynchrony during stimulation. Patterns of network spiking activity before, during and after training reproduce most of the in vivo physiological observations in the literature.

Selective delay activity in the cortex: phenomena and interpretation.

The present article does not intend to present technical progress nor recent successes in accounting for experiments, as this issue of the journal presents a rich inventory. Rather, the paper presents a retrospective reflection on the history of the subject; on the relation between the different aspects of the concepts and the phenomena involved; on its strengths and weaknesses; and on some future prospects. It is a tribute to an extremely rich and growing wealth of physiological phenomena and of interpretative concepts. Yet the extent of achievement is used to expose open questions, which appear to become ever deeper. It is also an attempt to make the subject a matter of discourse between biologists and modelers, without the distraction of technical details.

Spike-driven synaptic dynamics generating working memory states.

The collective behavior of a network, modeling a cortical module of spiking neurons connected by plastic synapses is studied. A detailed spike-driven synaptic dynamics is simulated in a large network of spiking neurons, implementing the full double dynamics of neurons and synapses. The repeated presentation of a set of external stimuli is shown to structure the network to the point of sustaining working memory (selective delay activity). When the synaptic dynamics is analyzed as a function of pre- and postsynaptic spike rates in functionally defined populations, it reveals a novel variation of the Hebbian plasticity paradigm: in any functional set of synapses between pairs of neurons (e.g., stimulated-stimulated, stimulated-delay, stimulated-spontaneous), there is a finite probability of potentiation as well as of depression. This leads to a saturation of potentiation or depression at the level of the ratio of the two probabilities. When one of the two probabilities is very high relative to the other, the familiar Hebbian mechanism is recovered. But where correlated working memory is formed, it prevents overlearning. Constraints relevant to the stability of the acquired synaptic structure and the regimes of global activity allowing for structuring are expressed in terms of the parameters describing the single-synapse dynamics. The synaptic dynamics is discussed in the light of experiments observing precise spike timing effects and related issues of biological plausibility.

Retrospective and prospective persistent activity induced by Hebbian learning in a recurrent cortical network.

Recordings from cells in the associative cortex of monkeys performing visual working memory tasks link persistent neuronal activity, long-term memory and associative memory. In particular, delayed pair-associate tasks have revealed neuronal correlates of long-term memory of associations between stimuli. Here, a recurrent cortical network model with Hebbian plastic synapses is subjected to the pair-associate protocol. In a first stage, learning leads to the appearance of delay activity, representing individual images ('retrospective' activity). As learning proceeds, the same learning mechanism uses retrospective delay activity together with choice stimulus activity to potentiate synapses connecting neural populations representing associated images. As a result, the neural population corresponding to the pair-associate of the image presented is activated prior to its visual stimulation ('prospective' activity). The probability of appearance of prospective activity is governed by the strength of the inter-population connections, which in turn depends on the frequency of pairings during training. The time course of the transitions from retrospective to prospective activity during the delay period is found to depend on the fraction of slow, N-methyl-d-aspartate-like receptors at excitatory synapses. For fast recurrent excitation, transitions are abrupt; slow recurrent excitation renders transitions gradual. Both scenarios lead to a gradual rise of delay activity when averaged over many trials, because of the stochastic nature of the transitions. The model reproduces most of the neuro-physiological data obtained during such tasks, makes experimentally testable predictions and demonstrates how persistent activity (working memory) brings about the learning of long-term associations.

Mean field and capacity in realistic networks of spiking neurons storing sparsely coded random memories.

Mean-field (MF) theory is extended to realistic networks of spiking neurons storing in synaptic couplings of randomly chosen stimuli of a given low coding level. The underlying synaptic matrix is the result of a generic, slow, long-term synaptic plasticity of two-state synapses, upon repeated presentation of the fixed set of the stimuli to be stored. The neural populations subtending the MF description are classified by the number of stimuli to which their neurons are responsive (multiplicity). This involves 2p + 1 populations for a network storing p memories. The computational complexity of the MF description is then significantly reduced by observing that at low coding levels (f), only a few populations remain relevant: the population of mean multiplicity - pf and those of multiplicity of order square root pf around the mean. The theory is used to produce (predict) bifurcation diagrams (the onset of selective delay activity and the rates in its various stationary states) and to compute the storage capacity of the network (the maximal number of single items used in training for each of which the network can sustain a persistent, selective activity state). This is done in various regions of the space of constitutive parameters for the neurons and for the learning process. The capacity is computed in MF versus potentiation amplitude, ratio of potentiation to depression probability and coding level f. The MF results compare well with recordings of delay activity rate distributions in simulations of the underlying microscopic network of 10,000 neurons.

The eye and the hand: neural mechanisms and network models for oculomanual coordination in parietal cortex.

The coordinated action of the eye and the hand is necessary for the successful performance of a large variety of motor tasks based on visual information. Although at the output level the neural control systems for the eye and the hand are largely segregated, in the parietal cortex of the macaque monkey there exist populations of neurons able to combine ocular and manual signals on the basis of their spatial congruence. An expression of this congruence is the clustering of eye- and hand-related preferred directions of these neurons into a restricted region of the workspace, defined as field of global tuning. This domain may represent a neural substrate for the early composition of commands for coordinated oculo-manual actions. Here we study two different prototypical network models integrating inputs about retinal target location, eye position and hand position. In the first one, we model the interaction of these different signals, as it occurs at the afferent level, in a feed-forward fashion. In the second model, we assume that recurrent interactions are responsible for their combination. Both models account surprisingly well for the experimentally observed global tuning fields of parietal neurons. When we compare them with the experimental findings, no significant difference emerges between the two. Experiments potentially able to discriminate between these models could be performed.