The functional anatomy of the MMN: A DCM study of the roving paradigm
Introduction
Novel events, or oddballs, embedded in a stream of repeated events, or standards, produce a distinct response that can be recorded non-invasively with electrophysiological techniques such as electroencephalography (EEG). The mismatch negativity (MMN) is one of the ERP components elicited by any discriminable violation in the acoustic regularity. The MMN is believed to be an index of automatic change detection governed by a pre-attentive sensory memory mechanism (Näätänen, 1990). Despite being the subject of much research, the mechanisms behind MMN generation remain a topic of much debate. Recently, we provided evidence that the mechanisms underlying the MMN can be considered within a hierarchical inference or predictive coding framework (Garrido et al., 2007). 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 encompasses two previous hypotheses that have been debated in the literature; the adaptation hypothesis (May et al., 1999, Jääskeläinen et al., 2004) and the model-adjustment hypothesis (Winkler et al., 1996, Näätänen and Winkler, 1999, Sussman and Winkler, 2001). While the latter allows for adaptation effects (which the authors refer to as refractoriness), the adaptation hypothesis precludes a prediction or model-dependent contribution to the MMN. Predictive coding entails both adjustments to a generative model of stimulus trains and adaptation due to the increasing precision of predictions. We tested the relative contributions of model-adjustment and adaptation by formulating them as network models with plastic changes in extrinsic (model-adjustment) and intrinsic (adaptation) connections. We show that both model-learning and adaptation contribute to the MMN, consistent with predictive coding or model based explanations (Friston et al., 2006, Winkler, 2007).
It has been suggested that stimulus repetition engenders an echoic memory trace, which compares preceding and current stimuli (Näätänen, 1992). The MMN increases with the number of repetitions of a standard stimulus and is believed to reflect the strength of this trace (Sams et al., 1993). Repetition in roving paradigms, which are characterised by sporadic changes in the frequency of a repeating tone, enhance a slow positive wave from 50 to 250 ms post-stimulus in the standard ERP; the repetition positivity (RP) (Baldeweg et al., 2004). Both RP and MMN increase with repetitions of standards, suggesting that these are the ERP correlates of sensory memory formation and are (the same) electrophysiological signatures of sensory learning (Haenschel et al., 2005).
In the predictive coding framework (see also Friston, 2005, Baldeweg, 2006), evoked responses, corresponding to prediction error, drive perceptual inference (within-trial) and changes in connectivity (between trials) so that prediction error is suppressed with learning. Previously, we used dynamic causal models (DCMs) to explore network models underlying mismatch or oddball responses (Garrido et al., under review). This was achieved by explaining differences in the ERP evoked by standard and deviant tones on the basis of plastic changes within a cortical network. DCM models every time-bin and every channel in a single analysis and attempts to explain differences in the evoked responses, including the MMN and other components such as the N1 (Näätänen et al., 2005), in terms of changes in connection strengths. Our previous study employed a classical oddball paradigm, which meant that any differences in between ERPs evoked by standard and deviant tones could have been driven by a stimulus-specific N1 response as well a pure MMN response. Here we used a roving paradigm, where there were no acoustic differences between the standard and deviant. This ensured that any differences could not be explained stimulus-specific differences in the N1 contribution and enabled us to model the MMN per se.
In short, the key contributions of this study are firstly, to demonstrate the validity of the mechanism for MMN generation that we have proposed previously (Garrido et al., under review), and secondly, to show that the ensuing mismatch responses are due to learning and not to stimuli differences per se.
Section snippets
Stimuli
We studied twelve healthy volunteers aged 24–34 (4 female). 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 resulted from few
Results
Learning the acoustic environment through stimulus repetition changes connectivity within and between hierarchically organised cortical areas. This analysis comprised three parts: (i) confirmation that there is a significant differential response (MMN) between the first and sixth tone presentation; (ii) model selection to identify the most likely number and hierarchical deployment of sources causing these responses and (iii) hypotheses or model testing to establish that the MMN is mediated by
Discussion
Using dynamic causal modelling, we have presented further evidence to suggest: (i) the mismatch negativity (MMN) is generated by self-organised interactions within a hierarchy of cortical sources (Garrido et al., 2007) and (ii) the MMN rests on plastic change in both extrinsic (between-source) and intrinsic (within source) connections (Garrido et al.; under review). Critically, these conclusions are consistent with previous analysis of a conventional oddball paradigm but can now be generalised
Acknowledgments
The Wellcome Trust and the Portuguese Foundation for Science and Technology funded this work. We thank David Bradbury for technical support and the volunteers for participating in this study.
All the software necessary to implement these analyses are available as part of the SPM academic freeware (http://www.fil.ion.ucl.ac.uk/spm).
References (43)
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.
A neural mass model for MEG/EEG: coupling and neuronal dynamics
NeuroImage
(2003) - 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.
Prefrontal cortex involvement in preattentive auditory deviance detection: neuroimaging and electrophysiological evidence
NeuroImage
(2003) Bayesian estimation of dynamical systems: an application to fMRI
NeuroImage
(2002)Learning and inference in the brain
Neural Netw.
(2003)- et al.
Dynamic causal modelling
NeuroImage
(2003) - et al.
Dynamic causal modelling of evoked potentials: a reproducibility study
NeuroImage
(2007)
Functional neural dynamics underlying auditory event-related N1 and N1 suppression response
NeuroImage
Short-term plasticity in auditory cognition
Trends Neurosci.
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
Electric brain response to sound repetition in humans: an index of long-term-memory — trace formation?
Neurosci. Lett.
Differential contribution of frontal and temporal cortices to auditory change detection: fMRI and ERP results
NeuroImage
Comparing dynamic causal models
NeuroImage
Probabilistic mapping and volume measurement of human primary auditory cortex
NeuroImage
Separate time behaviors of the temporal and frontal mismatch negativity sources
NeuroImage
Laminar origins and terminations of cortical connections of the occipital lobe in the rhesus monkey
Brain Res.
Auditory frequency discrimination and event-related potentials
Electroencephalogr. Clin. Neurophysiol.
Cited by (340)
Brain dynamics of mental workload in a multitasking context: Evidence from dynamic causal modeling
2024, Computers in Human BehaviorMismatch negativity and polygenic risk scores for schizophrenia and bipolar disorder
2024, Schizophrenia ResearchPredictions about prosody facilitate lexical access: Evidence from P50/N100 and MMN components
2023, International Journal of PsychophysiologyStorm on predictive brain: A neurocomputational account of ketamine antidepressant effect
2023, Neuroscience and Biobehavioral Reviews