Losing Control: Sleep Deprivation Impairs the Suppression of Unwanted Thoughts

Unwanted memories often enter conscious awareness when individuals confront reminders. People vary widely in their talents at suppressing such memory intrusions; however, the factors that govern suppression ability are poorly understood. We tested the hypothesis that successful memory control requires sleep. Following overnight sleep or total sleep deprivation, participants attempted to suppress intrusions of emotionally negative and neutral scenes when confronted with reminders. The sleep-deprived group experienced significantly more intrusions (unsuccessful suppressions) than the sleep group. Deficient control over intrusive thoughts had consequences: Whereas in rested participants suppression reduced behavioral and psychophysiological indices of negative affect for aversive memories, it had no such salutary effect for sleep-deprived participants. Our findings raise the possibility that sleep deprivation disrupts prefrontal control over medial temporal lobe structures that support memory and emotion. These data point to an important role of sleep disturbance in maintaining and exacerbating psychiatric conditions characterized by persistent, unwanted thoughts.

knowledge of the face-scene pairs across the overnight interval.
On each trial, participants viewed a single face, together with two scenes: one that was paired with the face and another which featured in the experiment but was not paired with this particular face. Participants were instructed to indicate which scene was paired with the face via key press within 5 s. We asked participants to make this response as quickly and accurately as possible. The trial terminated once a response had been provided or the time limit expired, before the next trial began.
Recognition accuracy was calculated as the proportion of face-scene associations that were correctly identified. Data were analysed using a 3 (TNT

Unwanted Thoughts
Methodological Details S1 Electrodermal activity was recorded using a BIOPAC MP36R data acquisition system and AcqKnowledge (ACQ) 4.4.1 software (sampling rate = 2KHz). During the affect evaluation tasks, E-Prime-triggered square pulse outputs were transmitted to the MP36R unit via a BIOPAC STP35A interface enabling precise alignment of each stimulus onset to the skin conductance response (SCR) data. Two BIOPAC EL507 disposable adhesive electrodes were attached to the fingertips of the index and middle fingers of the non-dominant hand. The data were imported and preprocessed using PsPM (version 4.0.2; Bach & Friston, 2013).
A unidirectional first-order Butterworth high-pass filter with cut-off frequency 0.05 Hz was used to filter the data to account for the change of baseline activity during the duration of the recording sessions. The time series averaged over corresponding trials for each experimental condition (e.g. 'No-Think', negative valence) were then extracted, for each subject, for each of the two sessions (pre-TNT and post-TNT; TNT = Think/No-Think). An average 'session-specific' skin conductance level (SCL) was computed for each subject for each session. SCL was the skin conductance value measured for the first second after the presentation of the stimuli, averaged across all conditions. This first 1 s period after stimulus presentation is widely considered to be the 'latency' period for event-related evoked SCRs (Bach, Flandin, Friston, & Dolan, 2010;Braithwaite, Watson, Robert, & Mickey, 2013;Lim et al., 1997). This 'baseline' SCL was then subtracted from the rest of the measured skin conductance activity, which was deemed to belong to a canonical evoked SCR, taken for the entirety of the time that the stimulus was presented on screen during a trial, after excluding the first second (i.e. 5.5 seconds). The area under the curve for each condition was then computed for each subject, for each of the two sessions. This is akin to an analysis approach shown previously for spontaneous skin conductance fluctuations, except here we have used it to analytically quantify event-related evoked SCRs . The average SCRs elicited by scenes in each TNT condition and valence category were used to calculate the difference in SCRs across sessions (dSCR; SCR post-TNT -SCR pre-TNT). The same analysis pipeline was used for 2 both the sleep and the sleep deprivation groups. Data from 2 participants were unavailable due to technical issues (sleep group n=1; sleep deprivation group n=1). Furthermore, we excluded data from 3 participants in the sleep group who were SCR non-responders. 1

Losing Control: Sleep Deprivation Impairs the Suppression of Unwanted Thoughts
Methodological Details S2 Electrocardiography (ECG) was recorded using a BIOPAC MP36R data acquisition system and AcqKnowledge (ACQ) 4.4.1 software (sampling rate = 2KHz). Three BIOPAC EL503 ECG electrodes were attached to the midline of the left and right clavicle and the lower left rib. ECG was recorded for eight successive minutes. The first 2 min and last 1 min of each recording was discarded, and heart rate variability (HRV) was calculated for the remaining five successive minutes. The ECG signal was analysed offline using ACQ and Kubios Standard 3.0.2 software.
R-peaks were automatically detected using ACQ's QRS detection algorithm and visually inspected for accuracy. Peaks that the algorithm missed were inserted manually. The interbeat-interval time series was then imported to Kubios for analysis.
Left and right electrooculogram, left, right and upper electromyogram, and a ground electrode (forehead) were also attached. All electrodes were verified to have a connection impedance of < 5 kΩ. All signals were digitally sampled at a rate of 200