Difficulty: Intermediate | Prerequisites: Basic knowledge of cognitive psychology, familiarity with working memory concepts.
Sternberg, Luria and Sheppes (2021) showed that people can reduce their desire to use social media by deliberately directing their thoughts toward neutral content (a technique called attentional distraction). They measured this both behaviourally (lower self-reported desire) and neurally (reduced EEG markers of working-memory representation and attention allocation). The core takeaway: loading working memory with unrelated, neutral thoughts limits how strongly social-media temptations take hold in your mind.
Self-regulation
The ability to control or modify one's own thoughts, emotions, desires and behaviours in the service of longer-term goals.
Think of it as your internal braking system: the thing that stops you opening Instagram when you should be revising.
Working memory (WM)
An online, limited-resource cognitive buffer that keeps relevant information in an active state and supports goal-directed behaviour (Engle, 2002).
In simple terms, working memory is the mental workspace where your brain holds whatever you are paying attention to right now. It can only hold a small amount, so whatever fills it tends to drive your next action.
Attentional distraction
A self-regulation strategy that involves deliberately loading working memory with neutral, unrelated content (e.g. thinking about geometric shapes or daily routines) so that tempting information cannot dominate the limited WM resource.
In simple terms, you crowd out the tempting thought by filling your head with something boring instead.
Contralateral delay activity (CDA)
A negative slow-wave ERP component recorded over posterior-parietal electrodes (PO7/PO8). CDA amplitude indexes how much information is actively maintained in visual working memory. Larger (more negative) CDA = more items held in WM.
Think of it as a volume dial for WM: the louder the signal, the more your brain is actively holding onto that content.
Late positive potential (LPP)
A positive slow-wave ERP component measured at centroparietal sites (Pz, CPz). LPP amplitude reflects the degree to which initial attention is allocated to affective or appetitive stimuli. Larger LPP = more attentional engagement.
In simple terms, a bigger LPP means your brain is paying more attention to the emotionally or motivationally grabbing stimulus.
Temptation condition
The experimental condition in which participants passively watched Facebook images and were instructed to let social-media-related thoughts and desires arise naturally, without trying to control them.
Attentional-distraction condition
The experimental condition in which participants viewed the same Facebook images but were instructed to redirect their thoughts toward neutral content (geometric shapes or daily routine activities) to control the influence of the social-media stimuli.
Event-related potential (ERP)
A measured brain-voltage response that is time-locked to a specific sensory, cognitive or motor event. ERPs are extracted from the ongoing EEG by averaging many trials together, which cancels out background noise and reveals the neural response to the event of interest.
Think of it as a snapshot of what your brain is doing at the exact moment something happens on screen.
Social media temptations compete with basic biological and safety needs. The study opens with statistics that frame the scale of the problem:
51.7% of teenagers regularly use electronic devices instead of going to sleep (Royant-Parola et al., 2018)
55% of users check their smartphone (primarily for social networking) before getting out of bed (Jilisha et al., 2019)
Up to 18% of fatal driving accidents are linked to mobile-phone usage (Overton et al., 2015)
53.5% of young drivers report regularly using Facebook at the wheel (Gauld et al., 2017)
Social-media updates are the main reason pedestrians give for using their phones while crossing the street (Byington & Schwebel, 2013)
The overarching point: social media does not just waste time. In extreme cases, the failure to regulate social-media urges can be life-threatening. This is why understanding whether and how these temptations can be self-regulated is a public-health concern, not just a productivity one.
Self-regulation of social media urges is known to be effortful and prone to breakdown, especially when the temptation is strong (Lyngs et al., 2019). Before this study, most research had examined social media as a tool people use to regulate other negative emotions (e.g. using Instagram to manage worry), rather than treating social-media content itself as the thing that needs regulating. Sternberg et al. (2021) set out to fill that gap directly.
The theoretical backbone of this study draws on canonical self-regulation models (Barrett et al., 2004; Hofmann & Van Dillen, 2018). The central claim is:
Working memory is a limited-capacity buffer that holds information in an active state.
When a tempting stimulus (e.g. a Facebook notification) enters WM, it can hijack those limited resources, directing attention toward action plans for giving in (e.g. opening the app, checking notifications).
The stronger the mental representation of the temptation in WM, the more likely a person is to act on it.
Therefore, if you can limit how much of that tempting content occupies WM, you weaken the temptation's grip on behaviour.
Attentional distraction works by loading WM with neutral content early, before the temptation has fully established itself. This crowds out the tempting representation and the associated consumption intentions.
Prior work had already shown attentional distraction to be effective against:
Food cravings (Van Dillen et al., 2013)
Processing of smoking-related cues (Littel & Franken, 2011)
Romantic feelings toward ex-partners (Langeslag & Sanchez, 2018)
Sexual desire (Shafir et al., 2018)
This study extends the same logic to social-media temptation, which had not been tested as a direct target of self-regulation before.
The study uses two ERP components to track what happens in the brain:
CDA tracks how much information is being held in visual WM. If attentional distraction works, CDA should be smaller (less negative) in the distraction condition, indicating that less social-media content is actively maintained.
LPP tracks how much initial attention is allocated to the tempting stimulus. If distraction works, LPP should be reduced, indicating less attentional engagement with the Facebook images.
Both measures give the study temporal precision that behavioural self-reports alone cannot provide.
N = 30 recruited; 1 excluded for excessive EEG artefacts (48% trial rejection rate). Final sample: 29 participants (22 female; mean age 25.07, SD = 2.70).
Sample size was predetermined via formal power analysis (paired-samples t-test, alpha = .05, 80% power, expected Cohen's d = 0.54 based on Sessa et al., 2011).
Inclusion criteria: normal or corrected-to-normal vision, normal colour vision, active daily Facebook user.
Participants had their Facebook password changed 24 hours before the experiment, blocking access for a total of 48 hours. This served two purposes:
It increased the value and saliency of Facebook stimuli during the task (a well-established deprivation technique used across many domains, including substance research).
It ensured participants did not use Facebook immediately before or after the EEG session.
Important: prior work (Sternberg et al., 2018) showed this deprivation does not distort natural usage patterns. Deprived lab usage time correlated positively with normal at-home usage time.
100 Facebook-related images were selected (10 for instruction, 90 for the experiment). Images contained recognisable Facebook elements: the Facebook icon, unread-notification icon, unread-message icon, friend-request icon, Facebook wall, and Messenger interface. Images were scaled to 3.3 x 4.5 degrees of visual angle at a 70 cm viewing distance, optimised for CDA analysis.
Each trial followed this sequence:
Fixation cross (200 ms)
Instruction screen: "Distraction" or "Watch" (250 ms)
Blink window (700 ms)
Arrow cue pointing left or right (200 ms), indicating which side of the screen to attend
Two Facebook stimuli presented bilaterally (one on each side of the screen) for 2,500 ms. Bilateral presentation is required for CDA analysis.
Desire rating: participants rated their current desire to use their own Facebook profile on a 1 (not feeling a desire at all) to 5 (feeling extreme desire) scale.
The task consisted of 16 blocks of 25 trials each, for a total of 400 analysed trials. Both conditions used identical stimuli, and trials with eye movements were excluded, ruling out low-level perceptual explanations for any differences.
Temptation condition ("Watch"): participants watched the Facebook images naturally, letting social-media-related thoughts and desires arise freely (e.g. thinking about their profile, recent posts, content they were missing).
Attentional-distraction condition ("Distraction"): participants directed their attention toward neutral, unrelated thoughts (geometric shapes or daily routine activities) while the same Facebook images were on screen.
Participants were trained and corrected by the experimenter before the task began. In distraction, they were corrected if their thoughts were not neutral, if they were related to Facebook, or if they did not fit the two approved categories. In temptation, they were corrected if they tried to control or regulate their thoughts.
60 randomly chosen trials included an on-screen check asking which instruction had just been used. Average accuracy: 91.66%.
Between blocks, the experimenter asked for verbal examples of how each instruction was implemented.
Participants were videotaped to ensure task concentration.
At the end, 8 additional trials required written descriptions of instruction implementation. A blind judge coded these as distraction or temptation with 96.5% accuracy.
64 scalp electrodes (extended 10-20 system) plus left and right mastoids.
EOG electrodes for detecting horizontal eye movements and blinks.
Digitised at 256 Hz. Processed with EEGLAB and ERPLAB toolboxes plus custom MATLAB scripts.
CDA measured at PO7/PO8, between 500 and 2,500 ms post-stimulus onset.
LPP measured at Pz and CPz, between 300 and 2,500 ms post-stimulus onset.
Mean trial rejection rate: 9.57% (CDA), 5.34% (LPP).
The most immediate practical application is teaching people to redirect attention from social-media cues in dangerous contexts. For instance, a driver who hears a notification sound could be trained to immediately focus on neutral visual elements of the road (signs, lane markings) rather than engaging with the thought of what the notification might contain. The same principle could be applied to students during study sessions or employees during focused work, turning a laboratory technique into a usable self-control habit.
Students often think attentional distraction means physically looking away from the screen or closing your eyes. It does not. Participants kept their gaze fixed at the centre; distraction here is entirely a top-down, mental strategy (redirecting thoughts, not redirecting gaze). Trials with eye movements were excluded precisely to rule this out.
Students sometimes confuse the CDA and LPP, treating them as interchangeable. They measure different things. CDA indexes how much content is being held in working memory. LPP indexes how much initial attention is being allocated to a stimulus. Both were reduced by distraction, but they capture different stages of processing.
It is a common error to assume this study proves attentional distraction is the best long-term strategy for managing social-media temptation. The authors explicitly note that distraction has long-term costs: it prevents habituation because the temptation is never fully processed. For sustained regulation, it may need to be combined with strategies like cognitive reappraisal.
Students sometimes think the Facebook deprivation was the intervention being tested. It was not. Deprivation was a methodological preparation step to increase the saliency of the stimuli. The intervention being tested was the attentional-distraction instruction during the EEG task.