Article Figures & Data
Figures
- Figure 1.
Experimental paradigm and examples of surprise models. a, Example of the task. Participants were trained on stimulus–response associations (top right) and then performed a simple choice-RT task. Visual stimuli were presented every 2 s for 200 ms each, requiring the participants to respond with the previously associated button. b, Raster plots of events sampled from two distributions where the probability of event 1 was either greater than all other events (high predictability environment) or the same (low-predictability environment). c–e, For each unique stimulus sequence models of blockwise stimulus-bound surprise (c), Ib, stimulus-bound surprise with no forgetting between blocks (d), Ie, and KL surprise (e).
- Figure 2.
Electrophysiological data. a, Scalp distribution (electrode Pz marked in cyan). b, Grand average waveform at electrode Pz. c, Average single-trial amplitude per stimulus category over midline electrode sites, showing effects traditionally reported for the P300. d, Results for two representative participants showing evoked potentials averaged to relative occurrence of stimuli each block (top right, 0.10 in blue, 0.25 in red, 0.40 in green, 0.70 in black) (top) and scatter plot of single-trial ERP amplitudes and blockwise surprise Ib (bottom). Scatter plot data were normalized for display purposes only. For ERPs, only the 200 ms before stimulus presentation (at time 0) to 800 ms following stimulus presentation interval is plotted, for display purposes only.
- Figure 3.
Model-based analysis of EEG data. a, Model comparison results comparing blockwise surprise Ib, to the traditional categorical model, KL surprise, and surprise without forgetting Ie. We report the log Bayes factor (or likelihood ratio), so that positive values > 3 indicate evidence (20:1 odds) in favor of Ib (indicated by dotted line). b, Posterior density of GLM parameters weighting the explanatory variable containing blockwise surprise at the group level. Error bar indicates SEM. c, Posterior densities showing influence of surprise on single-trial P300 estimate for each participant. Error bars indicate STD.
