A neuroanatomically grounded Hebbian‐learning model of attention–language interactions in the human brain

Abstract

Meaningful familiar stimuli and senseless unknown materials lead to different patterns of brain activation. A late major neurophysiological response indexing 'sense' is the negative component of event-related potential peaking at around 400 ms (N400), an event-related potential that emerges in attention-demanding tasks and is larger for senseless materials (e.g. meaningless pseudowords) than for matched meaningful stimuli (words). However, the mismatch negativity (latency 100-250 ms), an early automatic brain response elicited under distraction, is larger to words than to pseudowords, thus exhibiting the opposite pattern to that seen for the N400. So far, no theoretical account has been able to reconcile and explain these findings by means of a single, mechanistic neural model. We implemented a neuroanatomically grounded neural network model of the left perisylvian language cortex and simulated: (i) brain processes of early language acquisition and (ii) cortical responses to familiar word and senseless pseudoword stimuli. We found that variation of the area-specific inhibition (the model correlate of attention) modulated the simulated brain response to words and pseudowords, producing either an N400-or a mismatch negativity-like response depending on the amount of inhibition (i.e. available attentional resources). Our model: (i) provides a unifying explanatory account, at cortical level, of experimental observations that, so far, had not been given a coherent interpretation within a single framework; (ii) demonstrates the viability of purely Hebbian, associative learning in a multilayered neural network architecture; and (iii) makes clear predictions on the effects of attention on latency and magnitude of event-related potentials to lexical items. Such predictions have been confirmed by recent experimental evidence.

From sounds to words: A neurocomputational model of adaptation, inhibition and memory processes in auditory change detection

“…2C, and included the superior-temporal area model used in Experiment 1. Three inferior-frontal areas were added, with connectivity reflecting major features of the known neuroanatomical links within and between inferior frontal areas (see Methods, Experiment 1, and (Garagnani et al, 2008)). Neuroanatomical links between inferiorfrontal and superior-temporal areas were added to mimic long-distance connections by way of the extreme capsule and the arcuate fascicle, which have been documented to be present in macaques and even richer developed in humans (e.g., Pandya & Yeterian, 1985;Catani et al, 2005;Petrides & Pandya, 2009).…”

“…For nonlinguistic stimuli, there may be no clear MMN laterality or even laterality to the right. In earlier simulation studies, we have highlighted the role of frontotemporal memory circuits in explaining the spatiotemporal activation dynamics of frontal and temporal sources in language processing (Garagnani et al, 2008). Models solely relying on adaptation or local inhibition have difficulty explaining such specific effects in distant cortical areas.…”

“…An area-specific self-regulatory mechanism, termed "global" inhibition, was also implemented, in order to prevent the overall network activation from falling into non-physiological states (total saturation or inactivity). For further details of the model, see (Garagnani et al, 2008).…”

“…As described in detail by Garagnani et al (2008), the presence of Hebbian learning mechanisms in this architecture induces the emergence of model correlates of lexical representationscell assemblies (CAs) for words -as strongly connected distributed circuits that associate the paired "sensory" (A1) and "motor" (M1) activation patterns through neural elements located in areas linking auditory and motor cortex. The present network successfully learnt sensorimotor patterns for word pattern.…”

“…However, in view of the results of Experiment 2 and of previous simulations, we believe that this conclusion would not to be entirely accurate. In particular, we have previously shown that the presentation of not-represented stimuli (the model analogues of uncommon sounds, pseudowords, or other not previously learned items) to a model with memory circuits leads to the partial activation of such circuits (Garagnani et al, 2008;Garagnani, 2009). As mature brains are not tabula rasa entities but are equipped with a range of long-term memory traces, we conjecture that the same partial activations may occur in the brain when stimulated with pseudowords or other unfamiliar items (such as tone pips).…”