The research team recruited 61 participants and sorted them into three groups using well-established self-report instruments: the Social Interaction Anxiety Scale (SIAS), the Social Phobia Scale (SPS), and the Depression Anxiety and Stress Scale-21 (DASS-21). Nineteen participants formed the social anxiety group, twenty-one formed a group with elevated non-social anxiety symptoms, and twenty-one served as non-anxious controls. This three-way split matters because much previous research has lumped anxious individuals together, potentially masking meaningful differences between people whose worries center on social evaluation and those whose anxiety is diffuse or tied to non-social concerns. By separating the groups at the outset, the researchers could ask whether the same emotional attention machinery behaves differently across these profiles.

To probe that question, the team used a dual-task paradigm that combined an emotional Flanker task with a digit recognition task. The emotional Flanker task is a classic measure of emotional conflict regulation. Participants see a central target—typically a face expressing an emotion—flanked by distracting faces that are either congruent, meaning they express the same emotion, or incongruent, meaning they express a different one. The difference in performance between congruent and incongruent trials is called the emotional congruency effect, and a larger effect signals that the emotional content of the distractors is capturing attention and interfering with goal-directed responding. In this study, the flanking stimuli carried social evaluative weight, making the task especially relevant for people who fear scrutiny by others.

The digit recognition task served a second, equally important purpose: manipulating cognitive load. Under the low-load condition, participants held a small amount of digit information in working memory while performing the Flanker task; under the high-load condition, the memory demand was substantially greater. The logic follows from a long tradition in cognitive psychology showing that working memory resources are finite. When a demanding secondary task consumes those resources, less capacity remains for top-down attentional control, and emotionally salient distractors are free to exert whatever influence their bottom-up salience commands. The manipulation check was successful: reaction times for digit recognition were significantly longer and accuracy significantly lower under high load than under low load, confirming that the high-load condition genuinely taxed participants’ mental resources.

With the paradigm validated, the researchers turned to the emotional Flanker data, analyzed with repeated-measures analysis of variance across the three groups and the two load conditions. One of the earliest surprises came from raw reaction times. The social anxiety group responded significantly faster than both the elevated non-social anxiety group and the non-anxious group, a pattern the authors interpret as consistent with the heightened vigilance and rapid deployment of attention often described in social anxiety. Speed, however, is not the same as freedom from distraction, and the congruency effects told a more nuanced story about where each group’s attentional resources were actually going.

Under low cognitive load, both anxious groups showed significantly larger emotional congruency effects than the non-anxious controls. In other words, when their minds were relatively free, people in the social anxiety group and the elevated non-social anxiety group were both more distracted by the emotional mismatch between target and flankers than their non-anxious peers. This aligns with a large literature suggesting that anxiety amplifies the pull of threat-related and emotionally incongruent information, forcing the cognitive system to work harder to resolve conflict before an accurate response can be made. At low load, then, anxiety—whether social or non-social in character—appeared to leave a common signature of heightened emotional interference.

The picture changed dramatically when cognitive load was raised. Under high load, the three groups no longer differed significantly in their emotional congruency effects. For the elevated non-social anxiety group, the congruency effect shrank significantly under high load compared with low load, a reduction that reached the p < 0.01 threshold. The most parsimonious explanation is resource-based: the demanding digit task consumed so much working memory capacity that there was little left to elaborate or prioritize the emotional distractors, effectively flattening the anxiety-related amplification of emotional conflict. For these participants, a busy mind was, in a sense, a protected mind—overloaded cognition left no room for emotional distraction to exert its usual toll.

The social anxiety group defied this pattern. Their emotional congruency effects did not differ significantly between the high-load and low-load conditions, meaning that even when working memory was heavily taxed, emotional conflict continued to shape their performance much as it did when their minds were relatively free. This load-resistance suggests that the emotional attentional bias in social anxiety is not simply a byproduct of spare cognitive capacity that can be eliminated by occupying it. Instead, the authors argue, the bias may be more automatic, more deeply prioritized, or organized around the social evaluative content of the stimuli in a way that general cognitive load cannot easily suppress. For individuals who worry about being judged, emotionally incongruent social signals may command attention through channels that operate independently of deliberate, resource-dependent control.

The theoretical implications reach two influential frameworks. Attentional Control Theory proposes that anxiety impairs the efficiency of the goal-directed attentional system, particularly under pressure, while the Interaction of Emotion and Cognition Theory describes how emotional stimuli and cognitive processes compete for shared resources. The new findings refine the boundaries of both accounts by showing that these predictions hold cleanly for elevated non-social anxiety but not for social anxiety, at least when the distracting information is socially evaluative. A theory of anxious attention, the study implies, may need to specify not just how much load a person is under, but what kind of anxiety they carry and what kind of emotional content is competing for their mind.

The practical stakes are equally concrete. If the emotional interference in non-social anxiety can be reduced simply by loading cognition, then attentional training or task-design strategies that strategically occupy working memory might offer a route to quieter emotional distraction for those individuals. But the stability of the social anxiety effect across load conditions suggests that such load-based strategies may be less effective for socially anxious people, and that interventions should instead target the specific, load-resistant attentional priority given to social evaluative cues—perhaps through approaches that retrain the appraisal of social signals themselves. The authors note that their results provide empirical support for designing targeted attentional regulation interventions matched to distinct anxiety symptom profiles. The study was supported by a grant from the Science & Technology Development Fund of Tianjin Education Commission for Higher Education (2025SK140), received ethics approval from Tianjin Normal University, and was published open access on 28 September 2026. With 61 participants and a single experimental session, the findings await replication in larger and more diverse samples, but they sharpen a question that matters to anyone who has ever felt their attention hijacked by a worried mind: the answer, it turns out, depends on what kind of worry is doing the hijacking.

Subject of Research: The effect of cognitive load on emotional conflict attention in social anxiety versus non-social anxiety

Article Title: Cognitive Load and Emotional Flanker Performance in Individuals with Social Anxiety and Elevated Non-Social Anxiety Symptoms

Article References: Liu, C., & Hao, Z. (2026). Cognitive Load and Emotional Flanker Performance in Individuals with Social Anxiety and Elevated Non-Social Anxiety Symptoms. BMC Psychology. https://doi.org/10.1186/s40359-026-05642-8

Image Credits: AI Generated

DOI: 10.1186/s40359-026-05642-8

Keywords: cognitive load, social anxiety, emotional flanker task, dual-task paradigm, attentional control, emotional congruency effect, working memory, anxiety symptoms, emotion regulation, cognitive psychology, BMC Psychology, attentional bias

The technique in question is called episodic future thinking, a term first introduced by cognitive scientists Cristina Atance and Daniela O’Neill to describe the capacity to pre-experience events that have not yet happened. When a person mentally simulates a specific future episode, such as picturing themselves handing a document to a colleague in a particular office at a particular time, the brain recruits much of the same neural machinery used to recall genuine past experiences. Previous work, including studies by Mark Altgassen and colleagues and by Anja Kretschmer-Trendowicz and her collaborators, had already shown that episodic future thinking instructions improve prospective memory in older adults, adolescents and children. What remained contested was the mechanism: did imagining the future strengthen the memory for what had to be done, or did it sharpen the ability to notice the cue that triggers the action?

Prospective memory is conventionally decomposed into two separable components. The prospective component concerns noticing that the moment for action has arrived, detecting the target cue among a stream of ongoing activity. The retrospective component concerns retrieving what the intended action actually was once the cue has been noticed. These two processes can fail independently: a person may register that something was supposed to happen without recalling what, or recall the intention perfectly but never encounter or recognize the triggering cue. Because behavioral accuracy alone conflates the two, the Chinese research team employed a multinomial processing tree model, a mathematical framework originally developed by William Batchelder and Donald Riefer and adapted for event-based prospective memory by Rebecca Smith and Ute Bayen, which uses patterns of correct responses, errors and guesses to estimate the separate probabilities of the prospective and retrospective processes.

The researchers conducted two behavioral experiments with university students using a dual-task paradigm, in which participants performed an ongoing task while also holding an intention to respond to specific target cues. Experiment 1 used a between-subjects design with 112 participants, crossing two groups, an episodic future thinking group and a control group, with two levels of cue focality. Focal cues are ones that overlap with the processing demands of the ongoing task, so that the target is processed in the natural course of the ongoing activity, whereas non-focal cues require a shift of attention away from the ongoing task to be noticed. This focality distinction matters because prominent theories, notably the multiprocess framework of Mark McDaniel and Gilles Einstein, predict that non-focal intentions demand resource-demanding monitoring while focal intentions can benefit from relatively spontaneous retrieval.

The results of Experiment 1 were striking in their simplicity. Participants who had engaged in episodic future thinking before encoding the intention showed significantly higher prospective memory accuracy than controls, and crucially, this benefit emerged for both focal and non-focal cues. The advantage was not modulated by focality, suggesting that the strategy does not merely help in the easy, spontaneous-retrieval case but extends to situations where the cue is harder to catch. Equally informative was what did not change: episodic future thinking did not degrade performance on the ongoing task, and it did not alter the speed of prospective memory responses. In other words, the boost came free of charge, without the additional attentional cost that some monitoring-based strategies impose on whatever else a person is doing at the same time.

Experiment 2 pushed the logic further. If episodic future thinking works by strengthening the retrospective component, making the memory content itself more durable, then increasing the difficulty of that component should interact with the strategy’s benefit. The team tested 70 participants under the non-focal condition while deliberately increasing the difficulty of the retrospective component, requiring participants to hold and act on more demanding information about what to do when the cue appeared. The multinomial processing tree analysis delivered a decisive answer: even when the retrospective component was made harder, episodic future thinking significantly enhanced only the prospective component, the cue detection and monitoring process, and left the retrospective component statistically untouched.

This pattern of findings supports what the authors describe as a dual-pathway mechanism. Episodic future thinking appears to improve prospective memory through two complementary routes: it facilitates spontaneous retrieval, allowing the intention to pop into mind without deliberate search, and it strengthens controlled cue monitoring, the more effortful process of checking the environment for the triggering cue. The core effect, however, sits squarely on the monitoring side. By vividly simulating the future episode at encoding, a person apparently builds a richer, more accessible representation of the cue context, which lowers the threshold for noticing that cue when it later appears in the stream of ongoing activity. The intention is not better stored as content; it is better armed to be triggered.

The technical contribution of the study lies in its methodological rigor. Behavioral accuracy alone could never have distinguished between a monitoring account and a content-strength account, because both predict higher hit rates. By fitting a multinomial model to the full pattern of responses, including trials where participants correctly performed the ongoing task but missed the intention, the researchers could estimate latent process parameters that are not directly observable. The appendix parameters of the model, including probabilities of recognizing prospective memory targets, recognizing non-targets, and guessing, allowed the team to isolate exactly where in the cognitive architecture the improvement occurred. This modeling approach, increasingly common in prospective memory research, transforms a coarse accuracy difference into a mechanistic claim.

The practical implications reach well beyond the laboratory. Prospective memory failures are a hallmark of healthy aging, and they are profoundly disrupted in conditions such as Parkinson’s disease, attention deficit disorders and after brain injury. Prior trials, such as the randomized-controlled study by Stacey Goedeken and colleagues on encoding strategy training in Parkinson’s patients, hint that teaching people to encode intentions through future simulation could become a low-cost, widely applicable rehabilitation tool. The present findings refine the prescription: the strategy works because it primes cue monitoring, so it should be paired with interventions that make cues salient in the real environment, from phone notifications to strategically placed objects. The fact that the benefit carries no ongoing-task cost is particularly important for everyday multitasking, where remembering to remember often competes with driving, working or conversing.

The study also connects to a broader theoretical debate that has animated the field since Einstein and McDaniel’s foundational work in 1990. For decades, researchers argued over whether prospective memory is driven mainly by automatic, spontaneous processes or by effortful, capacity-consuming monitoring. The multiprocess view holds that both operate, with cue focality determining which dominates. The new evidence adds a twist: an encoding strategy can simultaneously boost both pathways, and its signature effect is on the monitoring process even for cues that should, in principle, be caught spontaneously. As future thinking research expands, with meta-analytic work by Gioia Cona and colleagues mapping the shared neural correlates of future simulation, prospective memory and delay discounting, the picture emerging is of a mind that remembers the future by first living it, and that the act of pre-living an intention quietly recalibrates the attentional systems that will later recognize the moment has come. The Henan team’s work, supported by the Henan Provincial Science and Technology Research Project, marks a substantial step toward making that recalibration deliberate, teachable and clinically useful.

Subject of Research: Cognitive mechanisms by which episodic future thinking enhances prospective memory encoding and retrieval

Article Title: Episodic future thinking as an effective encoding strategy for prospective memory: unraveling its underlying cognitive mechanisms

Article References: Guo, Y., Mao, J., Ma, F., Gan, J., & Li, Y. (2026). Episodic future thinking as an effective encoding strategy for prospective memory: unraveling its underlying cognitive mechanisms. Current Psychology, 45(19), Article 1571. https://doi.org/10.1007/s12144-026-10122-8

Image Credits: AI Generated

DOI: 10.1007/s12144-026-10122-8

Keywords: episodic future thinking, prospective memory, cue focality, spontaneous retrieval, cue monitoring, multinomial processing tree model, retrospective component, prospective component, dual-task paradigm, encoding strategy, cognitive psychology, memory