Distinctive Representation of Mispredicted and Unpredicted Prediction Errors in Human Electroencephalography

Hsu, Yi Fang; Bars, Solène Le; Hämäläinen, Jarmo A.; Waszak, Florian

doi:10.1523/jneurosci.2204-15.2015

Cited by 72 publications

(76 citation statements)

References 56 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Despite these limitations, our work provides further evidence for the effect of uncertainty-or precisionon prediction error, which is consistent with an increasing number of empirical findings (Garrido et al, 2013;Hsu et al, 2015;Lumaca et al, 2019;Sedley et al, 2016;Sohoglu & Chait, 2016;Southwell & Chait, 2018), theories of predictive processing and models of music perception (Clark, 2016;Feldman & Friston, 2010;Hohwy, 2013;Ross & Hansen, 2016;Vuust et al, 2018). Furthermore, our findings confirm that MMNm responses can be reliably recorded in realistic paradigms where sounds constantly change, which constitutes a methodological improvement on existing approaches.…”

Section: Limitations and Future Directionssupporting

confidence: 88%

“…The ensuing precision-weighted prediction error would ensure that primarily reliable sensory signals drive learning and behavior. While a growing body of research already provides evidence for this phenomenon in the auditory modality (Garrido, Sahani, & Dolan, 2013;Hsu, Bars, Hämäläinen, & Waszak, 2015;Lumaca, Haumann, Brattico, Grube, & Vuust, 2019;Sedley et al, 2016;Sohoglu & Chait, 2016;Southwell & Chait, 2018), our study was the first to show its presence in a more ecologically valid setting such as music listening. Furthermore, the findings also pointed to a feature-selective effect in which only prediction error responses related to the manipulated auditory feature-pitch, in our case-are modulated by uncertainty.…”

Section: Introductionmentioning

confidence: 69%

See 1 more Smart Citation

Musical prediction error responses similarly reduced by predictive uncertainty in musicians and non-musicians

Quiroga-Martinez¹,

Hansen

Højlund

et al. 2019

Preprint

View full text Add to dashboard Cite

Auditory prediction error responses elicited by surprising sounds can be reliably recorded with musical stimuli that are more complex and realistic than those typically employed in EEG or MEG oddball paradigms. However, these responses are reduced as the predictive uncertainty of the stimuli increases. In this study, we investigate whether this effect is modulated by musical expertise. Magnetic mismatch negativity (MMNm) responses were recorded from 26 musicians and 24 non-musicians while they listened to low-and high-uncertainty melodic sequences in a musical multi-feature paradigm that included pitch, slide, intensity, and timbre deviants. When compared to non-musicians, musically trained participants had significantly larger pitch and slide MMNm responses. However, both groups showed comparable reductions of pitch and slide MMNm amplitudes in the high-uncertainty condition compared to the lowuncertainty condition. In a separate, behavioral deviance detection experiment, musicians were more accurate and confident about their responses than non-musicians, but deviance detection in both groups was similarly affected by the uncertainty of the melodies. In both experiments, the interaction between uncertainty and expertise was not significant, suggesting that the effect is comparable in both groups. Consequently, our results replicate the modulatory effect of predictive uncertainty on prediction error; show that it is present across different types of listeners; and suggest that expertise-related and stimulus-driven modulations of predictive precision are dissociable and independent.

show abstract

Section: Limitations and Future Directionssupporting

confidence: 88%

Section: Introductionmentioning

confidence: 69%

Musical prediction error responses similarly reduced by predictive uncertainty in musicians and non-musicians

Quiroga-Martinez¹,

Hansen

Højlund

et al. 2019

Preprint

View full text Add to dashboard Cite

show abstract

“…Direct neural data (e.g., from intracranial recordings) might provide a more sensitive test of this proposal. Taken to an extreme (i.e., without any sensory input), computation of negative Prediction Errors could also explain previous results showing that the omission of an expected stimulus causes an increased signal [70–72] from which stimulus identity can be decoded [72–74]. …”

Section: Discussionmentioning

confidence: 72%

Prediction Errors but Not Sharpened Signals Simulate Multivoxel fMRI Patterns during Speech Perception

2016

View full text Add to dashboard Cite

Successful perception depends on combining sensory input with prior knowledge. However, the underlying mechanism by which these two sources of information are combined is unknown. In speech perception, as in other domains, two functionally distinct coding schemes have been proposed for how expectations influence representation of sensory evidence. Traditional models suggest that expected features of the speech input are enhanced or sharpened via interactive activation (Sharpened Signals). Conversely, Predictive Coding suggests that expected features are suppressed so that unexpected features of the speech input (Prediction Errors) are processed further. The present work is aimed at distinguishing between these two accounts of how prior knowledge influences speech perception. By combining behavioural, univariate, and multivariate fMRI measures of how sensory detail and prior expectations influence speech perception with computational modelling, we provide evidence in favour of Prediction Error computations. Increased sensory detail and informative expectations have additive behavioural and univariate neural effects because they both improve the accuracy of word report and reduce the BOLD signal in lateral temporal lobe regions. However, sensory detail and informative expectations have interacting effects on speech representations shown by multivariate fMRI in the posterior superior temporal sulcus. When prior knowledge was absent, increased sensory detail enhanced the amount of speech information measured in superior temporal multivoxel patterns, but with informative expectations, increased sensory detail reduced the amount of measured information. Computational simulations of Sharpened Signals and Prediction Errors during speech perception could both explain these behavioural and univariate fMRI observations. However, the multivariate fMRI observations were uniquely simulated by a Prediction Error and not a Sharpened Signal model. The interaction between prior expectation and sensory detail provides evidence for a Predictive Coding account of speech perception. Our work establishes methods that can be used to distinguish representations of Prediction Error and Sharpened Signals in other perceptual domains.

show abstract

“…Isochronous stimuli yield a reduced N1 compared to more complex, non‐isochronous rhythms (e.g. Clementz, Barber, & Dzau, ; Hsu, Le Bars, Hämäläinen, & Waszak, ; Lange, , ; Schafer, Amochaev, & Russell, ). This can be accounted for by attention‐related mechanisms.…”

Section: Discussionmentioning

confidence: 99%

Weighting of neural prediction error by rhythmic complexity: A predictive coding account using mismatch negativity

Lumaca

Haumann

Brattico

et al. 2019

Eur J of Neuroscience

View full text Add to dashboard Cite

The human brain's ability to extract and encode temporal regularities and to predict the timing of upcoming events is critical for music and speech perception. This work addresses how these mechanisms deal with different levels of temporal complexity, here the number of distinct durations in rhythmic patterns. We use electroencephalography (EEG) to relate the mismatch negativity (MMN), a proxy of neural prediction error, to a measure of information content of rhythmic sequences, the Shannon entropy. Within each of three conditions, participants listened to repeatedly presented standard rhythms of five tones (four inter‐onset intervals) and of a given level of entropy: zero (isochronous), medium entropy (two distinct interval durations), or high entropy (four distinct interval durations). Occasionally, the fourth tone was moved forward in time that is it occurred 100 ms (small deviation) or 300 ms early (large deviation). According to the predictive coding framework, high‐entropy stimuli are more difficult to model for the brain, resulting in less confident predictions and yielding smaller prediction errors for deviant sounds. Our results support this hypothesis, showing a gradual decrease in MMN amplitude as a function of entropy, but only for small timing deviants. For large timing deviants, in contrast, a modulation of activity in the opposite direction was observed for the earlier N1 component, known to also be sensitive to sudden changes in directed attention. Our results suggest the existence of a fine‐grained neural mechanism that weights neural prediction error to the complexity of rhythms and that mostly manifests in the absence of directed attention.

show abstract

Distinctive Representation of Mispredicted and Unpredicted Prediction Errors in Human Electroencephalography

Cited by 72 publications

References 56 publications

Musical prediction error responses similarly reduced by predictive uncertainty in musicians and non-musicians

Musical prediction error responses similarly reduced by predictive uncertainty in musicians and non-musicians

Prediction Errors but Not Sharpened Signals Simulate Multivoxel fMRI Patterns during Speech Perception

Weighting of neural prediction error by rhythmic complexity: A predictive coding account using mismatch negativity

Contact Info

Product

Resources

About