Limits...
Attention and prediction in human audition: a lesson from cognitive psychophysiology.

Schröger E, Marzecová A, SanMiguel I - Eur. J. Neurosci. (2015)

Bottom Line: Auditory attention typically yields enhanced brain activity, whereas auditory prediction often results in attenuated brain responses.As predictions encode contents and confidence in the sensory data, and as gain can be modulated by the intention of the listener and by the predictability of the input, various possibilities for interactions between attention and prediction can be unfolded.From this perspective, the traditional distinction between bottom-up/exogenous and top-down/endogenous driven attention can be revisited and the classic concepts of attentional gain and attentional trace can be integrated.

View Article: PubMed Central - PubMed

Affiliation: Institute for Psychology, BioCog - Cognitive and Biological Psychology, University of Leipzig, Neumarkt 9-19, D-04109, Leipzig, Germany.

Show MeSH

Related in: MedlinePlus

Illustration of the generation of auditory ERPs/ERP effects in the case of the occurrence of an unpredicted sound (left), a fully predicted sound (middle), and the omission of a strongly predicted sound (right). Left: the sensorial input arriving in the lower area is compared with the prediction sent down from the higher area. As there is no prediction for the arriving sound, the prediction error contains the complete input to the lower area. Middle: the sensorial input is met by a perfect prediction sent down from the higher area (e.g. when the sound is self-generated). As the input and prediction match perfectly, the prediction error is zero and no prediction error is sent to the higher area. The sensory input can be fully accounted for by the inferred causes (the model of predicted states) and the sound is perceived. Right: a prediction is sent down, but no input arrives at the lower area (as in the omission paradigm). The prediction error is then identical to the prediction. Thus, omission responses are a means to make the prediction (which is not easily accessible to the ERP technique) visible. This is a simplified version of figures published by Arnal & Giraud (2012) and SanMiguel et al. (2013c).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4402002&req=5

fig02: Illustration of the generation of auditory ERPs/ERP effects in the case of the occurrence of an unpredicted sound (left), a fully predicted sound (middle), and the omission of a strongly predicted sound (right). Left: the sensorial input arriving in the lower area is compared with the prediction sent down from the higher area. As there is no prediction for the arriving sound, the prediction error contains the complete input to the lower area. Middle: the sensorial input is met by a perfect prediction sent down from the higher area (e.g. when the sound is self-generated). As the input and prediction match perfectly, the prediction error is zero and no prediction error is sent to the higher area. The sensory input can be fully accounted for by the inferred causes (the model of predicted states) and the sound is perceived. Right: a prediction is sent down, but no input arrives at the lower area (as in the omission paradigm). The prediction error is then identical to the prediction. Thus, omission responses are a means to make the prediction (which is not easily accessible to the ERP technique) visible. This is a simplified version of figures published by Arnal & Giraud (2012) and SanMiguel et al. (2013c).

Mentions: Skeleton of the flow of information as postulated by the predictive coding theory. Generative models encoding the inferred causes of the sensorial input in representational units (R), located at a higher level, send predictions down to a lower level, where they are compared with the input arriving at the lower level from a still lower level (hierarchical system). The mismatch between the two is computed by prediction error (PE) units and sent forward to the higher level, so that the generative model can be improved and kept faithful. As only the PE is passed on to the higher levels, the amount of sensory data that need to be processed further is reduced to only those parts that are not already accounted for by the model. The system tries to minimize the PE, which is assumed to be generated by superficial pyramidal cells, where ERPs (and high-frequency oscillatory activity) are generated to a large extent. Deep pyramidal cells seem to be involved in the transmission of information backwards throughout the hierarchy, where more sustained ERPs (and low-frequency oscillatory activity) are generated. This is a simplified version of a figure published in Friston (2005) as Fig. 2.


Attention and prediction in human audition: a lesson from cognitive psychophysiology.

Schröger E, Marzecová A, SanMiguel I - Eur. J. Neurosci. (2015)

Illustration of the generation of auditory ERPs/ERP effects in the case of the occurrence of an unpredicted sound (left), a fully predicted sound (middle), and the omission of a strongly predicted sound (right). Left: the sensorial input arriving in the lower area is compared with the prediction sent down from the higher area. As there is no prediction for the arriving sound, the prediction error contains the complete input to the lower area. Middle: the sensorial input is met by a perfect prediction sent down from the higher area (e.g. when the sound is self-generated). As the input and prediction match perfectly, the prediction error is zero and no prediction error is sent to the higher area. The sensory input can be fully accounted for by the inferred causes (the model of predicted states) and the sound is perceived. Right: a prediction is sent down, but no input arrives at the lower area (as in the omission paradigm). The prediction error is then identical to the prediction. Thus, omission responses are a means to make the prediction (which is not easily accessible to the ERP technique) visible. This is a simplified version of figures published by Arnal & Giraud (2012) and SanMiguel et al. (2013c).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4402002&req=5

fig02: Illustration of the generation of auditory ERPs/ERP effects in the case of the occurrence of an unpredicted sound (left), a fully predicted sound (middle), and the omission of a strongly predicted sound (right). Left: the sensorial input arriving in the lower area is compared with the prediction sent down from the higher area. As there is no prediction for the arriving sound, the prediction error contains the complete input to the lower area. Middle: the sensorial input is met by a perfect prediction sent down from the higher area (e.g. when the sound is self-generated). As the input and prediction match perfectly, the prediction error is zero and no prediction error is sent to the higher area. The sensory input can be fully accounted for by the inferred causes (the model of predicted states) and the sound is perceived. Right: a prediction is sent down, but no input arrives at the lower area (as in the omission paradigm). The prediction error is then identical to the prediction. Thus, omission responses are a means to make the prediction (which is not easily accessible to the ERP technique) visible. This is a simplified version of figures published by Arnal & Giraud (2012) and SanMiguel et al. (2013c).
Mentions: Skeleton of the flow of information as postulated by the predictive coding theory. Generative models encoding the inferred causes of the sensorial input in representational units (R), located at a higher level, send predictions down to a lower level, where they are compared with the input arriving at the lower level from a still lower level (hierarchical system). The mismatch between the two is computed by prediction error (PE) units and sent forward to the higher level, so that the generative model can be improved and kept faithful. As only the PE is passed on to the higher levels, the amount of sensory data that need to be processed further is reduced to only those parts that are not already accounted for by the model. The system tries to minimize the PE, which is assumed to be generated by superficial pyramidal cells, where ERPs (and high-frequency oscillatory activity) are generated to a large extent. Deep pyramidal cells seem to be involved in the transmission of information backwards throughout the hierarchy, where more sustained ERPs (and low-frequency oscillatory activity) are generated. This is a simplified version of a figure published in Friston (2005) as Fig. 2.

Bottom Line: Auditory attention typically yields enhanced brain activity, whereas auditory prediction often results in attenuated brain responses.As predictions encode contents and confidence in the sensory data, and as gain can be modulated by the intention of the listener and by the predictability of the input, various possibilities for interactions between attention and prediction can be unfolded.From this perspective, the traditional distinction between bottom-up/exogenous and top-down/endogenous driven attention can be revisited and the classic concepts of attentional gain and attentional trace can be integrated.

View Article: PubMed Central - PubMed

Affiliation: Institute for Psychology, BioCog - Cognitive and Biological Psychology, University of Leipzig, Neumarkt 9-19, D-04109, Leipzig, Germany.

Show MeSH
Related in: MedlinePlus