Neural correlates of Predicting WHAT and WHEN in Visual Domain

Full text

Neural correlates of Predicting WHAT and WHEN in Visual Domain
Sanjeev Nara1, Asier Zarraga1,Mathieu Bourguignon1,3 ,Nicola Molinaro1,2
1Basque Center on Cognition, Brain and Language (BCBL). Donostia. Spain
2 Ikerbasque, Basque Foundation for Science, Spain
3ULB Neurosciences Institute, Université libre de Bruxelles, Belgium
INTRODUCTION RESULTS
References:
Rao and Ballard. Predicitve Coding in Visual Cortex, a functional interpration of some
extra classical receptive field effects,Nature Neuroscience, 79 -87 (1999)
Friston, K. A theory of cortical responses. Phil. Trans. R. Soc. B 360, 815–836 (2005).
Arnal et. Al. Cortical Oscillations and Sensory Predictions, Trends in Cog Sci. 390 – 397
(2012)
:
Discussion:
•When the timing was fixed, predictive trials elicited differential beta
power in a time range following the 4th entrainer up to 140 ms. The non-
predictive condition elicited more power in Beta range (25-30 Hz) which
could possibly reflect the lack of reliable prediction and the consequent
cost of estimating the visual properties of the target.
•When the timing is variable, Gamma power up to 260 miliseconds after
the fourth entrainer was larger when comparing Random trials with
Predictable trials.
•The present findings highlight the role of both beta and gamma
oscillatory activity in the development of predictive visual
representations.
•When timing is fixed, beta activity in posterior brain regions is
associated with the development of prediction.
•When timing is variable the additional involvement of left
prefrontal regions could reflect the stronger recruitment of top-
down resources
•Temporal preidictability is introduced by varying the Time Gap (G) between Entrainers
•Stimuli Variables : Non/Predictable Targets (target angle random/scaled), Orientation of
Target gabor (Vertical/Horizontal), High/Low CPD & Random/Scaled Timing
CONCLUSIONS
METHOD
Data:
Twenty right handed (7 males, age range 18 – 30, mean: 25.2 ,
SD: 3.3) healthy participants (no visual deficit) took part in the
study. Data were acquired using 306 channel MEG (Elekta,
Neuromag) with sampling frequency of 1000 Hz.
PreProcessing:
• Temporal signal source separation (tsss) Maxfiltered
•ICA, jump artefact rejection and muscle artifacts removed
•Time frequency estimation was performed using a Hanning
taper within a range of 1 to 70 Hz (1 Hz frequency resolution).
Introduction:
•Predictive Coding (Rao et. al 1999 ): the brain hosts an
internal representation of world (internal prior) which is
used to constantly estimate and evaluate the sensory
inputs.
•The timing of input stimuli (Predictive Timing) also
plays an important role in making these Predictions.
• In the present study we evaluate the interaction
between Predictive Coding and Predictive Timing in the
visual domain.
•Participants saw “contextual” gabor patches
(entrainers) whose orientation could predict / not
predict the target gabor orientation. Their timing was
fixed or variable. Participants task was to evaluate the
gabor thickness (Cycles Per Degree).
Predictable Trials Unpredictable Trials
Predicting WHEN(Cond_2) Predicting None (Cond_4)
Predicting WHAT + WHEN (Cond_1) Predicting WHAT (Cond_3)
Cond_2 VS Cond_1 (0 -350 msec) Cond_4 VS Cond_3 (0 -350msec)
Time (msec)
Time (msec)
Time (msec)
Time (msec)
Frequency (Hz)
Frequency (Hz)
Frequency (Hz)
Frequency (Hz)
Predictable Timing (Equal G) Unpredictable Timing (unequal G)