Repetition suppression and plasticity in the human brain
Introduction
We have previously used the roving paradigm and dynamic casual modelling (DCM) to search for optimum functional architectures underlying mismatch responses elicited by deviant and standard tones (Garrido et al., 2008). We were able to show that one can account for these responses with changes in connectivity among distributed cortical sources. In this paper, we adopt a parametric DCM to examine the form of repetition-dependent connectivity changes that mediate the emergence of these response differences. In brief, we attempt to model plasticity or changes in connectivity as a function of repetition or time. With this new approach we were able to quantify the time course of repetition-dependent changes and show that connectivity reduces, both within and between cortical areas. This causes decreases in evoked responses; i.e., repetition suppression, which manifests as a suppression of a mismatch responses, as an oddball becomes a standard. A key practical advance, afforded by this parametric DCM, is the ability to quantify the rate of experience-dependent plasticity non-invasively, using simple and established paradigms. Furthermore, because we use a physiologically informed model, one can assess plasticity separately in intrinsic and extrinsic connections. This may be useful in clinical and neuropharmacological studies.
Novel events, or oddballs, embedded in a stream of repeated events, or standards, produce a distinct response that can be recorded non-invasively with electroencephalography (EEG). For example, the mismatch negativity (MMN) is the negative component of the waveform obtained by subtracting the event-related response to a standard from the response to an oddball, or deviant event. The MMN is believed to index an automatic change detection by pre-attentive sensory memory mechanisms (Tiitinen et al., 1994). Recently, we provided evidence that the mechanisms underlying the MMN can be considered within a hierarchical inference or predictive coding framework (Garrido et al., 2007a, Garrido et al., 2007b). Within this account, the MMN is interpreted as a failure to suppress prediction error, which can be explained quantitatively in terms of coupling changes among and within cortical regions. The predictive coding framework accommodates two previous hypotheses; in the sense that it predicts the adjustment of a perceptual model of the current stimulus [c.f., the model-adjustment hypothesis (Winkler et al., 1996, Näätänen and Winkler, 1999)] and entails adaptive changes in post-synaptic sensitivity during learning [c.f., the adaptation hypothesis (May et al., 1999, Jääskeläinen et al., 2004)]. The model-adjustment hypothesis postulates that the MMN reflects on-line modifications of a perceptual model that is updated when the auditory input does not match model predictions. In this context, the MMN is regarded as a marker for error detection, caused by a deviation from a learned regularity. In other words, the MMN is a response to an unexpected stimulus change and, from the point of view of predictive coding, signals prediction error.
The mechanisms underlying suppression of prediction error are closely related to repetition suppression, in which “repeated experience with the same visual stimulus leads to both short and long-term suppression of neuronal responses in subpopulations of visual neurons” (Desimone, 1996). This is closely related to stimulus-specific adaptation (Pérez-González et al., 2005) in the auditory system, where “fast, highly stimulus-specific adaptation and slower plastic mechanisms work together to constantly adjust neuronal response properties to the statistics of the auditory scene” (Nelken, 2004). Repetition suppression is a ubiquitous phenomenon that speaks to both predictive coding (e.g., Friston, 2005) and models of perceptual inference and learning (e.g., Desimone, 1998). Predictive coding models of perceptual inference and learning suggest that all experience-dependent effects, and in particular repetition effects (from post-synaptic adaptation to semantic priming), are mediated by changes in synaptic efficacy; either short or long-term. These changes are driven by associative plasticity to optimise predictions of sensory input and therefore explain away prediction errors more efficiently. This enhanced ‘explaining away’ may be a useful perspective on repetition suppression, which rests on synaptic plasticity at the cellular level or changes in ensemble coupling at the macroscopic level. Critically, hierarchical inference, or predictive coding, also rests on optimising the relative influence of bottom-up prediction error and top-down predictions. This involves optimising the efficacy of intrinsic connections within an area or source (Friston, 2008). Put simply, when an unpredictable stimulus occurs, units encoding prediction error should adapt, reducing the strength of unreliable prediction error signals. In short, hierarchical inference, using prediction error, provides a principled framework in which the model adjustment and adaptation heuristics become necessary for understanding sensory inference (see Garrido et al., 2009a).
Few studies have explicitly explored the role of stimulus repetition during auditory memory-trace formation. Näätänen and Rinne (2002) found that later negative responses (> 100 ms), in contrast to earlier responses, are elicited only by sound repetition. Others found that increasing the number of repetitions enhances responses to standard tones in both early (30–50 ms) and later components (60–75 ms) (Dyson et al., 2005), localised in the primary and secondary auditory areas respectively (Liegeois-Chauvel et al., 1994). Similarly, Baldeweg et al., 2004, Haenschel et al., 2005 found that the MMN increases with the number of preceding standards and may be mediated by a repetition-dependent enhancement of a slow positive wave (50–250 ms) in the standard ERP (a repetition positivity — RP). In other words, the emergence of repetition positivity in standards underlies the mismatch negativity observed in a subsequent oddball.
Here we used a roving paradigm to test the hypothesis that repetition-dependent changes in electrophysiological responses to repeated stimuli are due to experience-dependent plasticity, or changes in connectivity. We show that stimulus repetition reduces connectivity, within and between cortical areas. This causes experience-dependent decreases in evoked responses; i.e., repetition suppression, which manifests as a suppression of MMN components, as an oddball becomes sufficiently predictable to be considered as a standard.
Section snippets
Subjects and stimuli
We studied ten healthy volunteers aged 24 to 34 (four females). Each subject gave signed informed consent before the study, which proceeded under local ethical committee guidelines. Subjects sat in front of a desk in a dimly illuminated room. Electroencephalographic activity was measured during an auditory roving ‘oddball’ paradigm (see Fig. 1a). The stimuli comprised a structured sequence of pure sinusoidal tones, with a roving, or sporadically changing tone. This paradigm entailed a few
Results
Our analysis comprised two parts: (i) confirmation that there is a significant response difference between the first and subsequent presentations of a tone and (ii) analysis of the plasticity in terms of coupling parameters that are a function of tone repetition.
Discussion
In summary, this study presents the first attempt to quantify plasticity underlying sensory memory formation, caused by stimulus repetition with a network of interacting cortical sources using EEG. We investigated the effect of stimulus repetition on scalp electroencephalographic responses and studied the underlying dynamics of the cortical network that generates these responses. Subjects were presented with a roving paradigm, a modified auditory oddball paradigm with tones that change
Acknowledgments
We thank David Bradbury for technical support and the volunteers for participating in this study. We thank Russell Poldrack for helpful comments on the manuscript. This work was supported by the Portuguese Foundation for Science and Technology (SFRH/BD/13481/2003 to M.I.G.) and the Wellcome Trust.
All the software necessary to implement these analyses is available as part of the SPM academic freeware (http://www.fil.ion.ucl.ac.uk/spm).
References (58)
Repetition effects to sounds: evidence for predictive coding in the auditory system
Trends Cogn. Sci.
(2006)- et al.
Mismatch negativity potentials and cognitive impairment in schizophrenia
Schizophr. Res.
(2004) - et al.
Evaluation of different measures of functional connectivity using a neural mass model
NeuroImage
(2004) - et al.
Modelling event-related responses in the brain
NeuroImage
(2005) - et al.
Dynamic causal modelling of evoked responses in EEG and MEG
NeuroImage
(2006) - et al.
Mechanisms of evoked and induced responses in MEG/EEG
NeuroImage
(2006) - et al.
Preictal short-term plasticity induced by intracerebral 1 Hz stimulation
NeuroImage
(2008) - et al.
Prefrontal cortex involvement in preattentive auditory deviance detection: neuroimaging and electrophysiological evidence
NeuroImage
(2003) - et al.
I’ve heard it all before: perceptual invariance represented by early cortical auditory-evoked responses
Brain Res. Cogn. Brain Res.
(2005) Bayesian estimation of dynamical systems: an application to fMRI
NeuroImage
(2002)
Learning and inference in the brain
Neural Netw.
Dynamic causal modelling
NeuroImage
Dynamic causal modelling of evoked potentials: a reproducibility study
NeuroImage
The functional anatomy of the MMN: a DCM study of the roving paradigm
NeuroImage
The mismatch negativity: a review of the underlying mechanisms
Clin. Neurophysiol.
Functional neural dynamics underlying auditory event-related N1 and N1 suppression response
NeuroImage
Dynamic causal modelling of evoked responses in EEG/MEG with lead field parameterization
NeuroImage
Dynamic causal modelling of evoked responses: the role of intrinsic connections
NeuroImage
Evoked potentials recorded from the auditory cortex in the man: evaluation and topography of the middle latency components
Electroencephalogr. Clin. Neurophysiol.
A neural mass model of spectral responses in electrophysiology
NeuroImage
Bayesian estimation of synaptic physiology from the spectral responses of neural masses
NeuroImage
Electric brain response to sound repetition in humans: an index of long-term-memory — trace formation?
Neurosci. Lett.
Processing of complex stimuli and natural scenes in the auditory cortex
Curr. Opin. Neurobiol.
Differential contribution of frontal and temporal cortices to auditory change detection: fMRI and ERP results
NeuroImage
Comparing dynamic causal models
NeuroImage
Multiple mechanisms of spike-frequency adaptation in motoneurones
J. Physiol. Paris
Probabilistic mapping and volume measurement of human primary auditory cortex
Neuroimage
Separate time behaviors of the temporal and frontal mismatch negativity sources
NeuroImage
Auditory frequency discrimination and event-related potentials
Electroenceph. Clin. Neurophys.
Cited by (171)
Changing connectivity between premotor and motor cortex changes inter-areal communication in the human brain
2023, Progress in NeurobiologyUnderstanding functional brain reorganization for naturalistic piano playing in novice pianists
2024, Journal of Neuroscience Research