Unreliable feedback deteriorates information processing in primary visual cortex

Description: In this fMRI study, we investigated how beliefs about sensory uncertainty influence early visual information processing. We used unreliable feedback interventions (in which invalid feedback was given on 50% of trials) to induce such beliefs and used reliable feedback as a control condition. The experiment used a between-group design, and fMRI data from 29 healthy human participants (n= 15 in the unreliable feedback group, 14 in the control/ reliable feedback group) were obtained. Participants performed a challenging visual orientation discrimination task by identifying visual gratings presented in circular annuli (inner r = 1.32°, outer r = 6.69°) as clockwise or counter-clockwise with reference to one of two implicit diagonal references (45° or 135°). The stimuli deviated from the diagonals by an angle defined prior to the main part of the experiment using a staircase procedure. The main part of the experiment had three phases (1) a baseline pre-intervention phase (no fb), (2) an intervention phase with reliable/unreliable fb and (3) a post-intervention phase (no fb). Here the following data are given for each participant: (1) Multivariate searchlight maps from pre-processed (slice-time correction, motion correction, normalisation) functional data. The maps show whole-brain estimates of "pattern distinctness", an index of multivariate representational precision (akin to the Mahalanobis Distance and as defined by Allefeld and Haynes, 2014, Neuroimage). The files are named as follows: *subject_number*test_phase*feedback_group*stimulus_pair_number where - subject_number ranges from 1 to 30. subject 22 was excluded due to excessive motion. - test_phase refers to pre and post-intervention phases (which were identical). - feedback group refers to the kind of feedback intervention used in the phase between pre and post-intervention. This can be reliable (control group) or unreliable (group of interest). - stimulus pair number: there were two pairs of stimuli - those that had the 45° or 135° diagonal as the implicit reference (i.e., the presented stimuli deviated by x° from the diagonals in a clockwise/counter-clockwise direction; x was defined for each participant using a staircase procedure prior to the experiment) (2) Univariate maps and V1 masks from pre-processed data, but in subject space. the maps show whole-brainn estimates of mean brain activity from the contrast. a The files were of two types: (1) showing mean activity in pre-/post-intervention test phases (all stimuli > 0). They were labelled as follows: con_0001*subject_number*test_phase*feedback_group* (2) V1 anatomical mask (voxels with > 50% probability, SPM anatomy toolbox, see Eickhoff et al., 2013). They were labelled as follows: anatMask*subject_number*test_phase*feedback_group*

Related article: http://doi.org/10.1016/j.neuroimage.2020.116701