Better performance for expected vs. unexpected working-memory tests.

a) Task schematic. Participants memorized two visual items, of which one would become tested for orientation reproduction. During the delay, a retrocue would indicate which memory item would most likely be tested. Across blocks, the retrocue was either 100%, 80%, or 60% reliable, or 100% reliable and imperative (no repeat of colour at test) as illustrated in the right panel. . b) Continuous reproduction error across four cue-reliability conditions (left panel), and for valid and invalid-cue trials in the 80% and 60% cue-reliability blocks (right panels). c) Response time across conditions. Error bars represent ±1SEM. *, **, n.s represent significance level p < 0.05, p < 0.01, and non-significant after Bonferroni correction.

Attentional orienting after the cue shows graded spatial modulations in both EEG-alpha activity and in gaze.

a) Time-frequency spectrum (contralateral minus ipsilateral) across frequency bands (2-40 Hz) and topographic distribution of 8-12 Hz EEG-alpha lateralization across conditions. b) Left panel: timeseries of alpha lateralization across conditions. Right panel: averages of alpha lateralization in a pre-defined time window (400-800 ms after cue, as in van Ede et al., 2019a) across conditions. c) Saccade rates separately for toward and away saccades across cue conditions relative to cue onset, colored lines represent saccades toward the memorized location of the cued item while grey lines represent saccades away from the memorized location of the cued item. On the side: spatial saccade bias (toward minus away, colourcoded) as a function of saccade size. d) Left panel: spatial saccade bias (toward minus away) across conditions. Right panel: averages of saccade bias in a pre-defined time window (200-600 ms after cue as in Liu et al., 2022) across conditions. Colored horizontal lines above x-axis in b,d indicate significant clusters. Shading and error bars represent ±1SEM. *, **, *** represent significance level p < 0.05, p < 0.01, and p < 0.001.

Attentional re-orienting after the memory test.

a) Alpha lateralization towards the memorized location of the tested item across conditions, here relative to after the onset of the memory test instead of the cue. From left to right panels, the alpha lateralization across imperative 100%, informative 100%, 80%, and 60% reliable-cue conditions. b) Saccade bias towards the memorized location of the tested item across conditions after the onset of the memory test. Colored horizontal lines above x-axis indicate significant clusters. The black horizontal lines indicate significant clusters comparing validly and invalidly cued tests. Plots at the bottom of panel b: onset and offset latency (defined by the 50% of the peak) of the spatial saccade bias following validly- and invalidly-cued memory tests in the 80% and 60% cue-reliability conditions. Circles and squares represent onset and offset latency, the light and dark colors represent validly- and invalidly-cued memory tests. Shading and error bars indicate ±1SEM. *, **, n.s represent significance level p < 0.05, p < 0.01, and not significant, respectively.

Re-orienting of attention after the memory test is not contingent on initial orienting after the cue.

a) Left panel: saccade bias towards the memorized location of the test item after memory test in trials with and without a toward saccade after the cue for validly cued trials in 80% cue-reliability block. Right panel: size of the test-locked saccade bias for the trials split by the preceding cue-locked saccades, averaged over the time window in which we observed the respective test-locked gaze bias in this condition. b) As per panel a, but for 60% valid-cue trials. c) and d), as per panels a and b, but for invalidly cued trials. The black horizontal line above the x-axis in panel d indicates the significant difference cluster between trials with and without a toward-saccade after the cue. Shading indicates ±1SEM.

Schematic summary of the main findings.

Schematic summary of our central findings regarding dynamic attentional orienting and re-orienting after the cue and the memory test, as a function of cue reliability and test expectations.