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

Abstract

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.

Integrating the Automatic and the Controlled: Strategies in Semantic Priming in an Attractor Network With Latching Dynamics

“…Indeed, such expectancy‐induced enhancement of the semantic feedback to the lexical layer has been suggested independently by several authors to account for other priming‐related phenomena (e.g., Brown, Stolz, & Besner, ; Robidoux, Stolz, & Besner, ; Stolz & Neely, ). Future developments of the model could examine whether such strengthening can be learned by a mechanism resembling the one used to adjust the noise in the present model, and clarify what are the experimental conditions which encourage such a strategy (see Lerner & Shriki, , for more information).…”

Semantic integration by pattern priming: experiment and cortical network model

“…If there was indeed no association between the primes and the target, these cases of''unrelated''priming would be incompatible with the fundamental process of propagation of activation involved in semantic priming (see Khalkhali et al 2012). However, we will see that the apparent absence of pair association and pair priming can be accounted for by a classical and simple mechanism reported at the neuronal level and used to describe activation in network models of priming (e.g., Anderson 1976Anderson , 1983aLavigne and Denis 2002;Mongillo et al 2003;Brunel and Lavigne 2009;Lerner et al 2012;Lerner and Shriki 2014;see McRae and Ross 2004;Randall et al 2004;Cree et al 1999;Becker et al 1997;Moss et al 1994;Masson et al 1991;Masson 1995;Plaut 1995;Plaut and Booth 2000). Indeed, in computational models, activation obeys a nonlinear, current-to-frequency transfer function inspired by the one observed in cortical neurons (Ricciardi 1977;Tuckwell 1988;Ermentrout and Kopell 1986).…”

“…Computational modeling allows linking behavioral data on semantic priming to biologically inspired properties of cortical networks (Mongillo et al 2003;Brunel and Lavigne 2009;Lavigne et al 2011;Lerner et al 2012;Lerner and Shriki 2014;see Bernacchia et al 2014). Results of the present experiment show that the processing of semantic patterns is content-specific at the level of triplets of words and not only of pairs of words.…”

Neuronal mechanisms for sequential activation of memory items: Dynamics and reliability

“…STD can clearly result in the inhibition of unit i. However in the framework of [1] it was not possible to obtain the spontaneous excitation of unit i + 2 with the connectivity matrix (3), because it was required that the upper and lower diagonal coefficients of J max be strictly increasing with the order i. Connectionist models have shown the effects of fast synaptic depression on semantic memory [78] and on priming [3] [4]. Fast synaptic depression contributes to deactivation of neurons initially active in a pattern -because they activate less and less each other -in favor of the activation of neurons active in a different but overlapping pattern -because 3 newly activated neurons can strongly activate their associates in a new pattern.…”