Mean error (in degrees). On each trial, participants must memorize three object orientations (e.g. 10, 40, and 75 degrees) that are shown at different locations in succession on a computer screen. Several seconds later, the participant is given a cue/indicator about which of the orientations they will need to recall (i.e., reproduce) during the trial's test phase. After another brief delay, a line is shown on the screen and the participant must rotate it with the mouse to match the orientation in memory. This occurred in two task conditions: the 'overlap' condition in which two memory items were shown at the same location on the screen; and the 'non-overlap' condition, in which all items were presented at different locations on the screen. The 'overlap' task was carried out in the first half of the measured time frame; the 'non-overlap' task was carried out in the second half. Large errors (big differences between the reported and displayed orientation) indicate worse memory accuracy.

The time, measured in milliseconds, that it took a subject to lock in their response via keyboard button press on each trial. The primary task completed by this group was a dual-serial recognition (DSR) task which involved two responses on each trial. The secondary (control) task was a single recognition (SR) task which involved one response on each trial. The two tasks were interleaved, so all participants completed them in succession throughout the measured time frame.

The average time, measured in milliseconds, that it took a subject to lock in their response via mouse button click on each trial. The primary task completed by this group of participants was a two-back recognition task, in which participants indicated whether the current stimulus was a match or non-match to the stimulus shown two items ago (completed during the first half of the measured time frame); the second task was a delayed recognition task in which participants indicated whether the current stimulus was a match or non-match to a stimulus shown 1750 ms prior (completed during the second half of the measured time frame).

The average time, measured in milliseconds, that it took a subject to lock in their response via keyboard button press on each trial. The primary task results were based on validly cued trials in which the cue indicated the to-be-tested location, whereas the second task results were based on the invalidly cued trials in which the cue indicated a location that was not the to-be-tested location. These task conditions were intermixed throughout the measured time frame.

The average time, measured in milliseconds, that it took a subject to lock in their response via keyboard button press on each trial. The primary task results were based on the accuracy of discrimination on 'congruent' trials in which the working memory item and the perceptual discrimination stimulus were identical in orientation; the secondary task results were based on the accuracy of discrimination on 'incongruent' trials. The two task types were intermixed throughout the measured time frame.

The average time, measured in milliseconds, that it took a subject to lock in their response via keyboard button press on each trial. The primary task was a 'set size 2' task in which participants had to maintain two items in working memory; the secondary task was a 'set size 1' task in which participants maintained one item in working memory. These tasks were intermixed throughout the measured time frame.

Multivariate pattern classification is a machine learning method that assesses the neural representation of stimulus information in electroencephalographic (EEG) signal (i.e., to "decode" the signal). The outcome measure is decoding performance. When a decoder performs well (here, greater than 0.5), the EEG signal contains information consistent with the representation of the stimulus at that point in the trial; when it performs poorly (here, less than or equal to 0.5), there is no evidence for stimulus representation at that time. Reported here is average classifier accuracy (expressed as area under the curve) in decoding a memory item's representation during the memory period for the different task conditions: when the item was cued or uncued and when transcranial magnetic stimulation (TMS) was delivered or not. The data used come from the full measured time frame.

Each EEG electrode's signal was construed as a weighted sum of responses from six channels tuned to specific stimulus orientations in the study. Data from the delayed recognition task were regressed onto this basis set to obtain a weight matrix characterizing the contribution of each channel to each electrode's response. The weight matrix was then inverted to derive the reconstructed representation of stimulus orientation during the delay period of the primary 2-back task data. Reported is the group-level slope (in arbitrary units) of the reconstruction of the cued and uncued memory items during the delay period, serving as a memory strength index/score ranging from -1:+1. Larger slope magnitudes indicate stronger memory representations. Negative values imply that the memory representation was a modified ('flipped') version of the representation held when the stimulus was first presented; positive values indicate direct resemblance to when the stimulus was first presented.

The spatially distributed phase coupling extraction (SPACE) decomposition method was used to identify discrete neural rhythms ('components') giving rise to the EEG signal. The strength of identified components at each time point (epoch) in the trial provides a measure of activity, ranging from 0 (absent/inactive) to a positive value (present/active). Strengths were used to address the question of whether single pulse TMS evokes new neural rhythms that weren't active before the pulse or modulates existing rhythms. If new rhythms are evoked, a greater percentage of trials with negligible (~0) strengths prior to TMS that increase after TMS compared to trials without TMS should be observed. Reported are the percentage of trials with this pattern for the posterior beta, posterior alpha, and posterior theta identified components for trials with and without TMS and TMS. Data from the full session were used.

Neural activity is comprised of rhythmic activity and aperiodic activity. Alpha rhythmic activity plays important roles in supporting working memory performance and varies according to task demands. EEG data were decomposed into alpha periodic and aperiodic components in order to isolate the alpha frequency band power (8-14 Hz). Participants' task was to make judgments about memorized items shown above, below, left, and right of a central viewing point on the screen. To assess the effects of spatial memory cueing on decomposed alpha, electrodes that showed alpha modulation selective to the four memory locations were first identified. Then, the alpha power in those electrodes was compared as a function of whether the location was attended, unattended, or irrelevant on a particular trial. This was done for two epochs: during the memory delay (350 - 850 ms after sample) and target presentation (850 - 1350 ms after sample). Data from the full session was used for this analysis.

The CDA is an event-related potential (ERP) derived from posterior electrodes that tracks the amount of information held in working memory, and may also be sensitive to context-binding demands. It becomes more negative with increasing memory load. The CDA was computed from the EEG by averaging the voltage across trials to generate signals that were contralateral or ipsilateral to the memory cue. The "difference wave" was computed by subtracting the ipsilateral signals from contralateral signals. The amplitude of the CDA is reported for two conditions: the large set size tracked participants' CDA when the memory set was homogenous - comprised of multiple items from the same stimulus category thus all were cued; the small set size analyses tracked participants' performance when the memory set was heterogenous - comprised of items from multiple stimulus categories, with only one category being cued for the target response. Data from the full session was used in this analysis.

The CDA is an event-related potential (ERP) derived from posterior electrodes that tracks the amount of information held in working memory, and may also be sensitive to context-binding demands. It becomes more negative with increasing memory load. The CDA was computed from the EEG by averaging the voltage across trials to generate signals that were contralateral or ipsilateral to the memory cue. The "difference wave" was computed by subtracting the ipsilateral signals from contralateral signals. The amplitude of the CDA is reported for two conditions: The large set size tracked participants' CDA when the memory set was comprised of two target features; the small set size analyses tracked participants' performance when the memory set was comprised of one target feature. Data from the full session was used in this analysis.

Multivariate inverted encoding modeling will be used to reconstruct the representation of stimulus locations from the electroencephalography data, and the strength of the representation will be compared across three stimulus conditions. Note that this method entails analysis of the broadband electroencephalographic signal (bandpass filtered from 1-100Hz) in each of two formats: time domain, and spectrally transformed. The spectrally transformed analysis does not entail the separate analysis of discrete functionally defined frequency bands (e.g., alpha, beta, etc.). Rather, spectral power values at every integer frequency from 2 to 20 Hz and at every other integer frequency from 22 to 50 Hz, yielding 34 frequencies per channel, are used as features in the analysis.

Spatially distributed phase coupling extraction-identified components of the transcranial magnetic stimulation-evoked electroencephalography signal will indicate whether the unattended memory item reactivation effect is carried by a de novo component in the electroencephalographic signal, or by a change in the power of one or more components that were present in the signal prior to the delivery of transcranial magnetic stimulation. Note that this method entails analysis of a spectral transformation of the broadband electroencephalographic signal that does not entail the separate analysis of discrete functionally defined frequency bands (e.g., alpha, beta, etc.) Rather, spectral power values at every integer frequency from 2 to 20 Hz and every other integer from frequency from 22 to 30 Hz - yielding 24 frequencies per channel - are entered into the analysis. No a priori assumptions are made about the frequency composition of components that the method will identify.

Correlation of the amplitude of multivariate inverted encoding model-reconstructions of the location of the unattended memory item, derived from the transcranial magnetic stimulation-evoked response, with alpha band power when targeting occipital cortex.

Correlation of the amplitude of multivariate inverted encoding model-reconstructions of the location of the unattended memory item, derived from the transcranial magnetic stimulation-evoked response, with beta-band power when targeting the intraparietal sulcus.

Frequency in the alpha band of the EEG as a function of retinotopic location

Spatially distributed phase coupling extraction-identified components of the electroencephalography signal from signals corresponding to the attended location to assess whether expectation-related shifts in alpha-band frequency are produced by a change in the frequency of one oscillator or by a change in the relative power of multiple oscillators.

Reaction time assess as latency to press response button after onset of critical stimulus.

Power in the alpha band of the EEG as a function of retinotopic location

Frequency in the alpha band of the EEG as a function of retinotopic location

Spatially distributed phase coupling extraction-identified components of the electroencephalography signal from signals corresponding to the attended location to assess whether expectation-related shifts in alpha-band frequency are produced by a change in the frequency of one oscillator or by a change in the relative power of multiple oscillators.

Multivariate inverted encoding modeling of the EEG signal to determine whether or not contextual information is carried in this signal. Note that this method entails analysis of the broadband electroencephalographic signal (bandpass filtered from 1-100Hz) in each of two formats: time domain, and spectrally transformed. The spectrally transformed analysis does not entail the separate analysis of discrete functionally defined frequency bands (e.g., alpha, beta, etc.). Rather, spectral power values at every integer frequency from 2 to 20 Hz and at every other integer frequency from 22 to 50 Hz - yielding 34 frequencies per channel -- are used as features in the analysis.

Posner-style visual cueing task involving orientation judgments; n = 19

Visual working memory task with three categories of stimuli: orientations, luminances, colors, and involving 'match'/'non-match' responses with EEG recorded; n = 28

Dual task with delayed recall of orientation (working memory) and visual discrimination of orientation during the delay period (perception) with EEG recorded; n = 25

Visual working memory task with delayed recognition responses ('match'/'non-match') of object color or orientation with EEG recorded; n = 56

Delayed-recall of orientation working memory task with EEG recorded; n = 14

Total of all reporting groups

Baseline characteristic of 'female' and 'male' is based on voluntary participant reports. In some instances, participants have opted not to disclose this information and as a result, the number analyzed in a particular sub-study will be different from the overall total enrolled.

Mean percentage of correct recognition responses - indicates the percentage of trials, on average, participants correctly identified a test stimulus as a match or non-match to an item in working memory. The primary task completed by this group of participants was a two-back recognition task, in which participants indicated whether the current stimulus was a match or non-match to the stimulus shown two items ago (completed during the first half of the measured time frame); the second task was a delayed recognition task in which participants indicated whether the current stimulus was a match or non-match to a stimulus shown 1750 ms prior (completed during the second half of the measured time frame).

EEG recorded; sub-study 2 participants only

Mean percentage of correct recognition responses - indicates the percentage of trials, on average, participants correctly identified a test stimulus as a match or non-match to an item in working memory. The primary task results were based on validly cued trials in which the cue indicated the to-be-tested location, whereas the second task results were based on the invalidly cued trials in which the cue indicated a location that was not the to-be-tested location. These task conditions were intermixed throughout the measured time frame.

EEG recorded; sub-study 3 participants only

Mean percentage of correct recognition responses - indicates the percentage of trials, on average, participants correctly identified a test stimulus as a match or non-match to an item in working memory. The primary task results were based on the accuracy of discrimination on 'congruent' trials in which the working memory item and the perceptual discrimination stimulus were identical in orientation; the secondary task results were based on the accuracy of discrimination on 'incongruent' trials. The two task types were intermixed throughout the measured time frame.

EEG recorded; sub-study 5 participants only

Mean error (in degrees). On each trial, participants must memorize three object orientations (e.g. 10, 40, and 75 degrees) that are shown at different locations in succession on a computer screen. Several seconds later, the participant is given a cue/indicator about which of the orientations they will need to recall (i.e., reproduce) during the trial's test phase. After another brief delay, a line is shown on the screen and the participant must rotate it with the mouse to match the orientation in memory. This occurred in two task conditions: the 'overlap' condition in which two memory items were shown at the same location on the screen; and the 'non-overlap' condition, in which all items were presented at different locations on the screen. The 'overlap' task was carried out in the first half of the measured time frame; the 'non-overlap' task was carried out in the second half. Large errors (big differences between the reported and displayed orientation) indicate worse memory accuracy.

EEG recorded; sub-study 7 participants only

The time, measured in milliseconds, that it took a subject to lock in their response via keyboard button press on each trial. The primary task completed by this group was a dual-serial recognition (DSR) task which involved two responses on each trial. The secondary (control) task was a single recognition (SR) task which involved one response on each trial. The two tasks were interleaved, so all participants completed them in succession throughout the measured time frame.

EEG recorded and single-pulse TMS applied intermittently during delay periods. Sub-study 1 participants only

The average time, measured in milliseconds, that it took a subject to lock in their response via mouse button click on each trial. The primary task completed by this group of participants was a two-back recognition task, in which participants indicated whether the current stimulus was a match or non-match to the stimulus shown two items ago (completed during the first half of the measured time frame); the second task was a delayed recognition task in which participants indicated whether the current stimulus was a match or non-match to a stimulus shown 1750 ms prior (completed during the second half of the measured time frame).

EEG recorded; sub-study 2 participants only

The average time, measured in milliseconds, that it took a subject to lock in their response via keyboard button press on each trial. The primary task results were based on validly cued trials in which the cue indicated the to-be-tested location, whereas the second task results were based on the invalidly cued trials in which the cue indicated a location that was not the to-be-tested location. These task conditions were intermixed throughout the measured time frame.

EEG recorded; sub-study 3 participants only

The average time, measured in milliseconds, that it took a subject to lock in their response via keyboard button press on each trial. The primary task results were based on the accuracy of discrimination on 'congruent' trials in which the working memory item and the perceptual discrimination stimulus were identical in orientation; the secondary task results were based on the accuracy of discrimination on 'incongruent' trials. The two task types were intermixed throughout the measured time frame.

EEG recorded; sub-study 5 participants only

The average time, measured in milliseconds, that it took a subject to lock in their response via keyboard button press on each trial. The primary task was a 'set size 2' task in which participants had to maintain two items in working memory; the secondary task was a 'set size 1' task in which participants maintained one item in working memory. These tasks were intermixed throughout the measured time frame.

EEG recorded; sub-study 6 participants only

Multivariate pattern classification is a machine learning method that assesses the neural representation of stimulus information in electroencephalographic (EEG) signal (i.e., to "decode" the signal). The outcome measure is decoding performance. When a decoder performs well (here, greater than 0.5), the EEG signal contains information consistent with the representation of the stimulus at that point in the trial; when it performs poorly (here, less than or equal to 0.5), there is no evidence for stimulus representation at that time. Reported here is average classifier accuracy (expressed as area under the curve) in decoding a memory item's representation during the memory period for the different task conditions: when the item was cued or uncued and when transcranial magnetic stimulation (TMS) was delivered or not. The data used come from the full measured time frame.

EEG recorded and single-pulse TMS applied intermittently during delay periods. Sub-study 1 participants only.

Each EEG electrode's signal was construed as a weighted sum of responses from six channels tuned to specific stimulus orientations in the study. Data from the delayed recognition task were regressed onto this basis set to obtain a weight matrix characterizing the contribution of each channel to each electrode's response. The weight matrix was then inverted to derive the reconstructed representation of stimulus orientation during the delay period of the primary 2-back task data. Reported is the group-level slope (in arbitrary units) of the reconstruction of the cued and uncued memory items during the delay period, serving as a memory strength index/score ranging from -1:+1. Larger slope magnitudes indicate stronger memory representations. Negative values imply that the memory representation was a modified ('flipped') version of the representation held when the stimulus was first presented; positive values indicate direct resemblance to when the stimulus was first presented.

EEG recorded; sub-study 2 participants only

The spatially distributed phase coupling extraction (SPACE) decomposition method was used to identify discrete neural rhythms ('components') giving rise to the EEG signal. The strength of identified components at each time point (epoch) in the trial provides a measure of activity, ranging from 0 (absent/inactive) to a positive value (present/active). Strengths were used to address the question of whether single pulse TMS evokes new neural rhythms that weren't active before the pulse or modulates existing rhythms. If new rhythms are evoked, a greater percentage of trials with negligible (~0) strengths prior to TMS that increase after TMS compared to trials without TMS should be observed. Reported are the percentage of trials with this pattern for the posterior beta, posterior alpha, and posterior theta identified components for trials with and without TMS and TMS. Data from the full session were used.

EEG recorded and single-pulse TMS applied intermittently during delay periods; sub-study 1 participants only

Neural activity is comprised of rhythmic activity and aperiodic activity. Alpha rhythmic activity plays important roles in supporting working memory performance and varies according to task demands. EEG data were decomposed into alpha periodic and aperiodic components in order to isolate the alpha frequency band power (8-14 Hz). Participants' task was to make judgments about memorized items shown above, below, left, and right of a central viewing point on the screen. To assess the effects of spatial memory cueing on decomposed alpha, electrodes that showed alpha modulation selective to the four memory locations were first identified. Then, the alpha power in those electrodes was compared as a function of whether the location was attended, unattended, or irrelevant on a particular trial. This was done for two epochs: during the memory delay (350 - 850 ms after sample) and target presentation (850 - 1350 ms after sample). Data from the full session was used for this analysis.

EEG recorded; Sub-study 3 participants only

The CDA is an event-related potential (ERP) derived from posterior electrodes that tracks the amount of information held in working memory, and may also be sensitive to context-binding demands. It becomes more negative with increasing memory load. The CDA was computed from the EEG by averaging the voltage across trials to generate signals that were contralateral or ipsilateral to the memory cue. The "difference wave" was computed by subtracting the ipsilateral signals from contralateral signals. The amplitude of the CDA is reported for two conditions: the large set size tracked participants' CDA when the memory set was homogenous - comprised of multiple items from the same stimulus category thus all were cued; the small set size analyses tracked participants' performance when the memory set was heterogenous - comprised of items from multiple stimulus categories, with only one category being cued for the target response. Data from the full session was used in this analysis.

with EEG recorded; Sub-study 4 participants only

The CDA is an event-related potential (ERP) derived from posterior electrodes that tracks the amount of information held in working memory, and may also be sensitive to context-binding demands. It becomes more negative with increasing memory load. The CDA was computed from the EEG by averaging the voltage across trials to generate signals that were contralateral or ipsilateral to the memory cue. The "difference wave" was computed by subtracting the ipsilateral signals from contralateral signals. The amplitude of the CDA is reported for two conditions: The large set size tracked participants' CDA when the memory set was comprised of two target features; the small set size analyses tracked participants' performance when the memory set was comprised of one target feature. Data from the full session was used in this analysis.

EEG recorded; sub-study 6 participants only

Multivariate inverted encoding modeling will be used to reconstruct the representation of stimulus locations from the electroencephalography data, and the strength of the representation will be compared across three stimulus conditions. Note that this method entails analysis of the broadband electroencephalographic signal (bandpass filtered from 1-100Hz) in each of two formats: time domain, and spectrally transformed. The spectrally transformed analysis does not entail the separate analysis of discrete functionally defined frequency bands (e.g., alpha, beta, etc.). Rather, spectral power values at every integer frequency from 2 to 20 Hz and at every other integer frequency from 22 to 50 Hz, yielding 34 frequencies per channel, are used as features in the analysis.

No data were collected for this outcome measure.

Spatially distributed phase coupling extraction-identified components of the transcranial magnetic stimulation-evoked electroencephalography signal will indicate whether the unattended memory item reactivation effect is carried by a de novo component in the electroencephalographic signal, or by a change in the power of one or more components that were present in the signal prior to the delivery of transcranial magnetic stimulation. Note that this method entails analysis of a spectral transformation of the broadband electroencephalographic signal that does not entail the separate analysis of discrete functionally defined frequency bands (e.g., alpha, beta, etc.) Rather, spectral power values at every integer frequency from 2 to 20 Hz and every other integer from frequency from 22 to 30 Hz - yielding 24 frequencies per channel - are entered into the analysis. No a priori assumptions are made about the frequency composition of components that the method will identify.

No data were collected for this outcome measure.

Correlation of the amplitude of multivariate inverted encoding model-reconstructions of the location of the unattended memory item, derived from the transcranial magnetic stimulation-evoked response, with alpha band power when targeting occipital cortex.

No data were collected for this outcome measure.

Correlation of the amplitude of multivariate inverted encoding model-reconstructions of the location of the unattended memory item, derived from the transcranial magnetic stimulation-evoked response, with beta-band power when targeting the intraparietal sulcus.

No data were collected for this outcome measure.

Frequency in the alpha band of the EEG as a function of retinotopic location

No data were collected for this outcome measure.

Spatially distributed phase coupling extraction-identified components of the electroencephalography signal from signals corresponding to the attended location to assess whether expectation-related shifts in alpha-band frequency are produced by a change in the frequency of one oscillator or by a change in the relative power of multiple oscillators.

No data were collected for this outcome measure.

Reaction time assess as latency to press response button after onset of critical stimulus.

Data were not collected for this outcome measure.

Power in the alpha band of the EEG as a function of retinotopic location

data were not collected for this outcome measure.

Frequency in the alpha band of the EEG as a function of retinotopic location

data were not collected for this outcome measure.

Spatially distributed phase coupling extraction-identified components of the electroencephalography signal from signals corresponding to the attended location to assess whether expectation-related shifts in alpha-band frequency are produced by a change in the frequency of one oscillator or by a change in the relative power of multiple oscillators.

data were not collected for this outcome measure

Multivariate inverted encoding modeling of the EEG signal to determine whether or not contextual information is carried in this signal. Note that this method entails analysis of the broadband electroencephalographic signal (bandpass filtered from 1-100Hz) in each of two formats: time domain, and spectrally transformed. The spectrally transformed analysis does not entail the separate analysis of discrete functionally defined frequency bands (e.g., alpha, beta, etc.). Rather, spectral power values at every integer frequency from 2 to 20 Hz and at every other integer frequency from 22 to 50 Hz - yielding 34 frequencies per channel -- are used as features in the analysis.

data were not collected for this outcome measure

2-back working memory task with EEG recorded involving 'match'/'non-match' responses; n = 30

Posner-style visual cueing task involving orientation judgments; n = 19

Visual working memory task with three categories of stimuli: orientations, luminances, colors, and involving 'match'/'non-match' responses with EEG recorded; n = 28

Dual task with delayed recall of orientation (working memory) and visual discrimination of orientation during the delay period (perception) with EEG recorded; n = 25

Visual working memory task with delayed recognition responses ('match'/'non-match') of object color or orientation with EEG recorded; n = 56

Delayed-recall of orientation working memory task with EEG recorded; n = 14