Example of the reference-back paradigm, including fMRI timing, as used in the experiment. In each trial, participants are tasked with indicating whether the probe stimulus (’X’ or ’O’) matches or differs from the stimulus presented in the most recent red frame, which serves as the working memory referent. In reference trials (red frame), participants are required to update their working memory with the currently displayed item. On the other hand, in comparison trials (blue frame), participants make the ’same/different’ decision but do not update their working memory.

Contrast weights for defining the four distinct working memory updating subprocesses.

Hierarchical descriptive statistics for the behavioral data.

The figure illustrates the mean response times (and standard error of the mean) in relation to two factors: a) whether the condition was switched or repeated, and b) whether the stimulus/response matched the previous reference stimulus or differed from it. The figure also displays the associated behavioral contrasts and their significance levels. See 2.2. for detailed information on how the contrasts were computed. Ref = reference trials; Com = comparison trials; GO = gate opening; GC = gate closing; S = substitution; UM = updating mode. *** = p < 0.001

Statistical parametric map of the ‘gate opening’ contrast with a threshold determined using the FWER method (p<.05; corresponding to a threshold of z = 2.3).

List of peak activation in MNI coordinates from the whole-brain analysis.

Statistical parametric map of the ‘gate closing’ contrast with a threshold determined using the FWER method (p<.05; corresponding to a threshold of z = 2.3).

Statistical parametric map of the ‘substitution’ contrast with a threshold determined using the FWER method (p<.05; corresponding to a threshold of z = 2.3).

Statistical parametric map of the ‘updating mode’ contrast with a threshold determined using the FWER method (p<.05; corresponding to a threshold of z = 2.3).

Subcortical statistical parametric mapping of the four contrasts with a focus on the midbrain (top row), and midbrain and thalamic regions (bottom row) using a FWER threshold of 2.3. Coordinates are in MNI2009cAsym (1mm) space. For orientation purposes, the MASSP atlas is overlaid: SN (blue), VTA (black), red nucleus (purple), subthalamic nucleus (green), periaqueductal grey (lightblue), GPi (yellow), GPe (orange), striatum (white), and thalamus (grey).

Results of the region-of-interest wise GLMs using the individually parcellated masks derived from MASSP. Error bars represent the 95% credible intervals. Abbreviations indicate thalamus (Tha), striatum (Str), globus pallidus externa (GPe), globus pallidus interna (GPi), subthalamic nucleus (STN), substantia nigra (SN), and ventral tegmental area (VTA). * indicate moderate evidence and ** indicate strong evidence that the observed BOLD signal change is larger than 0.

Results of the region-of-interest wise GLMs using the probabilistic atlas from Pauli, et al. (2018). Error bars represent the 95% credible intervals. Abbreviations indicate ventral tegmental area nucleus (VTA), parabrachial pigmented nucleus (PBP), and substantia nigra pars compacta (SNc). * indicate moderate evidence and ** indicate strong evidence that the observed BOLD signal change is larger than 0.

Results from the Bayesian one-sample t-test on the beta values derived from the ROI-wise GLMs are reported for each contrast and ROI. Moderate or higher evidence for ROI activity is indicated by bold font. Bayes factors favoring the alternative hypothesis are reported under BF10, and Bayes factors favoring the null hypothesis in column BF01.

Results from the Bayesian one-sample t-test on the beta values derived from the ROI GLMs using the masks from Pauli et al. (2018) are reported for each contrast and ROI. Moderate or higher evidence for ROI activity is indicated by bold font. Bayes factors favoring the alternative hypothesis are reported under BF10, and Bayes factors favoring the null hypothesis in column BF01.