Priming and memory: Behavior, Neurophysiology and Modeling

Editors

Frédéric Lavigne

University of Nice Sophia Antipolis

Laurent Dumercy

University of Nice Sophia Antipolis

Gianluigi Mongillo

Centre National de la Recherche Scientifique (CNRS)

Impact

Priming effect as measured in monkeys performing a symmetry decision task. After a fixation period (200 ms), either a symmetric (A) or an asymmetric (B) stimulus appeared on the screen, together with two cues. Monkeys were required to make a saccade upwards/downwards in response to a symmetric/antisymmetric stimulus. Priming was measured for both stimuli repetitions across different lags (C). Priming effect–as the difference between reaction times in response to the first and the second presentation of a stimulus- was larger when no other stimulus was interleaved between the two presentations, i.e., 0 lag (D). Adapted from McMahon and Olson (2007).

Review

22 January 2015

On the relationship between persistent delay activity, repetition enhancement and priming

Elisa M. Tartaglia

, 1 more and

Nicolas Brunel

6,647 views

17 citations

Probability of the types of dendrites (A–C) and synaptic efficacy after inter-synaptic learning (D–I), as a function of the number of synapses from neurons coding for the context and neurons coding for the stimulus (case of dendrites of neurons coding for response R1). For clarity, synapses from NSI-coding neurons are not reported (see Figure 1D for details). Insets focus on the probability or efficacy (y-axe) along the anti-diagonal of each figure, where the number of synapses from the combined items is constant (seven synapses). The x-axe (black anti-diagonal lines) corresponds to the number of synapses from the Stimulus (the inverse number of synapses from the Context). The effects of the increasing number of co-active synapses (from the context and from the stimulus) appear along the diagonal, while the effects of the increasing (decreasing) number of synapses from the stimulus (context) appear on the shape of the curve along the anti-diagonal (insets). Non-flat shapes reveal the effects of IS learning. (A–C) After random connectivity: heterogeneous distribution of the probability of the types of dendrite, defined by the numbers of synapses from pre-synaptic neurons coding for contexts (C1 or C2), stimuli (S1 or S2) and responses (R1 and R2): [C1, S1, C2, S2, R1, R2]. To better display slight variations along the scale, only dendrites occurring with a probability higher than 1.28 × 10−9 are displayed (note the different scales between graphs A, B and C). For a given dendrite, this corresponds to a maximum of seven synapses from pre-synaptic neurons coding for C1 and/or S1. (A) Overall probability (over all synapses from C2, S2, R1, R2, and NSI-coding neurons) of the types of dendrites, as a function of the number of synapses from neurons coding for C1 and S1 (mean = 0.02; min = 2.80 × 10−9; max = 0.11). (B,C) Two types of relevant dendrites (see text). (B) Overall probability of [C1 = 4, S1 = 4, C2, S2, R1, R2] dendrites (averaged over all types) with any number of synapses from C2, S2, R1, and R, as a function of the number of synapses from neurons coding for C2 and S2 (mean = 3.52 × 10−5; min = 7.50 × 10−10; max = 2.35 × 10−4). (C) Same as (B) but regarding [C1 = 2, S1 = 2, C2, S2, R1, R2] dendrites (mean = 0.001; min = 9.10 × 10−10; max = 8.65 × 10−3). (D-F) Amplification of potentiation and depression through IS learning of the combination C1S1R1: Sum of the efficacy of two synapses JR1−S + JR1−C from pre-synaptic neurons coding for the stimulus and the context for the different types of dendrites defined in (A–C). Amplification of potentiation and depression with the number of synapses in the dendrites indicates that IS learning of the C1S1R1 combination generates different efficacy values in different dendrite. (D) Amplification of the potentiation along the diagonal when the number of synapses increases (mean = 0.83, min = 0 and max = 1.22, compare to I). Synaptic efficacy is weak when one type of synapses (from C1 or S1) is absent along the anti-diagonal (inset). (E,F) Two types of dendrites having [4C1, 4S1] or [2C1, 2S1] synapses further show the efficacy of two synapses JR1−S1 + JR1−C1 as a function of the number of synapses from C2 and S2. Overall amplification of potentiation is stronger in [C1 = 4, S1 = 4, C2, S2, R1, R2] dendrites (E; mean = 0.89; min = 0.76 max = 1.17) than in [C1 = 2, S1 = 2, C2, S2, R1, R2] dendrites (F; mean = 0.67; min = 0.42; max = 1.08). In both cases, amplification of depression increases with the number of synapses from C2 and S2, also learned in combination with R1 but generating depression of the C1R1 and S1R1 synapses when learning a C2S2R1 combination. (G,H) Efficacy JR1C1 of synapses from C1 learned in combination with S1 and R1 (learned context, G) or of JR1C2 of synapses from C2 not learned in combination with S1 and R1 (context change, H). Overall synaptic efficacy from neurons coding for the context is stronger when the context was learned in the same combination (C1, G; mean = 0.41; min = 0; max = 0.61) than when it was not (C2, H; mean = 0.27; min = 0; max = 0.44). (I) Classical Hebbian learning of the combination C1S1R1: As in (D): synaptic efficacy is strong (mean = 0.58). However, the efficacy does not depend on the number of synapses in the dendrite along the diagonal, due to the absence of inter-synaptic amplification of potentiation (min = max = 0.67). The efficacy is weak when one type of synapses (from C1 or S1 or both) is absent along the anti-diagonal (inset).

Original Research

28 August 2014

Inter-synaptic learning of combination rules in a cortical network model

Frédéric Lavigne

, 1 more and

Laurent Dumercy

4,180 views

13 citations

General Commentary

11 August 2014

A latch on priming

Alberto Bernacchia

, 1 more and

Frédéric Lavigne

3,445 views

1 citations

Original Research

16 April 2014

Internally- and externally-driven network transitions as a basis for automatic and strategic processes in semantic priming: theory and experimental validation

Itamar Lerner

and

Oren Shriki

For the last four decades, semantic priming—the facilitation in recognition of a target word when it follows the presentation of a semantically related prime word—has been a central topic in research of human cognitive processing. Studies have drawn a complex picture of findings which demonstrated the sensitivity of this priming effect to a unique combination of variables, including, but not limited to, the type of relatedness between primes and targets, the prime-target Stimulus Onset Asynchrony (SOA), the relatedness proportion (RP) in the stimuli list and the specific task subjects are required to perform. Automatic processes depending on the activation patterns of semantic representations in memory and controlled strategies adapted by individuals when attempting to maximize their recognition performance have both been implicated in contributing to the results. Lately, we have published a new model of semantic priming that addresses the majority of these findings within one conceptual framework. In our model, semantic memory is depicted as an attractor neural network in which stochastic transitions from one stored pattern to another are continually taking place due to synaptic depression mechanisms. We have shown how such transitions, in combination with a reinforcement-learning rule that adjusts their pace, resemble the classic automatic and controlled processes involved in semantic priming and account for a great number of the findings in the literature. Here, we review the core findings of our model and present new simulations that show how similar principles of parameter-adjustments could account for additional data not addressed in our previous studies, such as the relation between expectancy and inhibition in priming, target frequency and target degradation effects. Finally, we describe two human experiments that validate several key predictions of the model.

5,310 views

26 citations

Original Research

14 April 2014

Tree shrews (Tupaia belangeri) exhibit novelty preference in the novel location memory task with 24-h retention periods

Jayakrishnan Nair

, 3 more and

Gregor Rainer

Top view of the experimental setup, illustrating the relative size of the objects and the tree shrew during exploration of one object.

Novelty preference is pervasive in mammalian species, and describes an inherent tendency to preferentially explore novelty. The novel location memory task studied here assesses the ability of animals to form accurate memories of a spatial configuration, consisting of several identical objects placed within an arena. Tree shrews were first familiarized with a particular object configuration during several sessions, and then an object was displaced during a test session. Tree shrews exhibited enhanced exploration when confronted with this novel configuration. The most reliable indicator associated with novelty preference was an enhancement in directed exploration towards the novel object, although we also observed a non-specific overall increase in exploration in one experiment. During the test session, we also observed an exploration of the location, which had previously been occupied by the displaced object, an effect termed empty quadrant. Our behavioral findings suggest multiple stages of spatial memory formation in tree shrews that are associated with various forms of behavioral responses to novelty. Reduced novelty preference has been linked to major depressive disorder in human patients. Given the established social conflict depression model in tree shrews, we anticipate that the study of the neural circuits of novelty preference and their malfunction during depression may have implications for understanding or treating depression in humans.

4,761 views

17 citations