Microexpressions: Can Brief Facial Cues Reveal Emotions?
Author: Angela Bogdanova · Published: October 13, 2026 · Editorial Policy
Research cluster: Emotion & Affect Science · Written by Artificial Sapiens
Microexpressions are very brief facial movements that researchers study as possible signals of emotion-related activity. They can be measured, and some people can learn to identify patterns presented briefly on a screen. Yet a fleeting movement does not directly disclose a person's inner emotional experience, motives, or honesty. Scientific studies provide mixed findings about whether microexpressions help distinguish truthful from deceptive behavior; current evidence does not justify treating them as a reliable lie detector. The distinction between observing a facial movement and interpreting what it means is central to the research. Barrett and colleagues' comprehensive review and a review of nonverbal deception research provide the broader scientific context.
This article examines what counts as a microexpression, how researchers record it, why emotional leakage has been proposed, what individual experiments actually found, and what training and automated analysis can and cannot establish. Its focus is the psychological evidence concerning facial movements. The broader interpretation of emotion from faces, voices, and context belongs to our emotion recognition overview.
What Is a Microexpression in Psychology?
A microexpression is commonly defined as a relatively rapid, transient configuration of facial movements, sometimes associated with an attempt to conceal or manage an emotional display. Researchers have used several thresholds. A narrow traditional definition covers roughly 1/25 to 1/5 of a second, or approximately 40 to 200 milliseconds. Other studies examine facial expressions lasting up to 300, 400, or 500 milliseconds. Consequently, a claim about how often microexpressions occur or what they predict cannot be evaluated without knowing how duration was defined. Porter and ten Brinke (2008) used a narrow interval; Matsumoto and Hwang (2018) explicitly compared broader duration cutoffs.
The word expression can imply that the face communicates a particular emotion. In research, the directly observable event is a configuration or sequence of movements: a change around the eyebrows, eyelids, mouth, or other facial regions. Its psychological meaning requires additional evidence. A twitch, a brief smile, a scowl, and a facial movement associated with discomfort do not all belong to the same psychological category. Speed alone never establishes emotional significance.
Microexpressions, macroexpressions, and subtle expressions
Microexpressions are categorized by their brevity. Longer or ordinary facial expressions may involve movements that unfold across a second or several seconds, although no universally accepted duration divides every type. Subtle expressions are defined primarily by low intensity or partial appearance, rather than by duration: a faint tightening around the eyes could be subtle without being especially brief. A blink, habit, speech-related movement, motor tic, or reaction to a physical stimulus may also be brief while having a different origin. Observers and algorithms must separate the event to be measured from the interpretation they want to make.
Why do definitions disagree?
Different studies pursue different tasks. A perceptual experiment may flash posed photographs for 40, 120, or 200 milliseconds to test whether viewers recognize a labeled configuration. An observational study may measure a complete movement from onset through its peak and recovery. A deception experiment may count any movement below a predefined threshold during an interview. These procedures yield different events and different denominators, so their reported accuracy rates are not interchangeable. In Shen, Wu, and Fu's experiments, recognition improved with longer presentation, reaching a leveling pattern around 200 milliseconds. That result concerns recognizing displayed images, not establishing the true emotion of their subjects.
Where Did the Idea of Emotional Leakage Come From?
The influential emotional-leakage hypothesis proposes that people sometimes attempt to hide an emotion while elements of a corresponding facial movement emerge involuntarily for a moment. A person might try to maintain a polite expression while briefly changing the eyebrows or mouth. The hypothesis links emotional processes, expressive control, and observation, but each link requires independent testing. The presence of a small facial movement is observable; the proposed concealed feeling is an inference; a claim about deception requires still another inference about the truthfulness of a statement.
Concealment, suppression, social politeness, and deliberate deception can overlap, but they are distinct. Someone may suppress laughter during a ceremony, conceal disappointment to protect another person, or remain expressionless in an uncomfortable setting. A person can also say something false without experiencing a strong emotion, or experience anxiety while saying something true. Because several mechanisms can generate similar facial behavior, a short-lived expression is not a unique signature of lying. Vrij, Hartwig, and Granhag's scientific review evaluates the weakness of behavioral cues to deception.
Does an unconscious facial signal reveal a hidden truth?
An observer can notice behavior that the person did not intend to display. That does not establish unconscious motivation or privileged access to the person's thoughts. Emotional experience, appraisal, attention, physiological change, and expression are related processes whose coordination changes with context. A fleeting facial action may be involuntary, partially controlled, habitual, socially communicative, or unrelated to a discrete emotion. Without independent measures of the target's experience, researchers cannot label the movement as a direct readout of a concealed feeling. The distinction reflects the broader evidence on the variability of human emotional expression.
How Do Scientists Observe and Measure Microexpressions?
High-frame-rate video and onset-to-offset timing
Video analysis allows researchers to inspect the beginning, peak, and conclusion of a facial action. Higher temporal resolution gives more observations within a short event, while lighting, camera position, occlusion, speaking, and head movement affect what can be measured. Researchers need explicit coding rules and independent raters or validated automated methods to decide whether a movement occurred. Measuring a 100-millisecond event from ordinary video is methodologically demanding because there may be only a small number of frames in which the movement is visible.
Timing also matters because an image flashed for 100 milliseconds is not equivalent to a naturally unfolding facial change that lasts 100 milliseconds. In the first case, the researcher controls visual exposure. In the second, the researcher measures behavior as it occurs. Confusing the two turns findings about visual recognition into unsupported claims about emotional leakage.
Facial Action Coding System and action units
The Facial Action Coding System, or FACS, offers a descriptive framework for coding visible facial movements into action units associated with muscular actions. Coders can describe a brow movement or lip movement without assuming a unique feeling caused it. This makes movement coding more defensible than assigning an emotion label at the first step. Still, coding reliability depends on the quality of the footage, operational definitions, training, and disagreements between raters. A precisely coded action unit is evidence of an observable movement; it is not a validated measurement of honesty or an internal emotional state. Barrett et al. (2019) review the inferential gap between expressive movement and emotion.
What should a good microexpression study report?
A persuasive study specifies the duration window, event start and end rules, frame rate, coder blinding, interrater reliability, how many people and expressions were sampled, whether emotion labels were independently verified, and how truthfulness was established. It should report the number of positive detections as well as missed events and false alarms. An apparent increase in average negative expressions across groups may be informative scientifically while remaining insufficient for individual decisions. Independent replication, preregistered thresholds, held-out test data, and realistic base rates strengthen the evidence.
Can Microexpressions Reveal Genuine Emotions?
Brief facial actions sometimes accompany emotional episodes, and some facial patterns carry information useful in social perception. Researchers can also reliably train participants to recognize specific briefly displayed configurations. Those findings justify studying microexpressions as one component of human emotional behavior. They do not make facial patterns perfect labels for happiness, anger, fear, shame, disgust, or any other subjective state.
The influential 2019 review by Barrett, Adolphs, Marsella, Martinez, and Pollak examined the assumptions behind inferring specific emotions from facial configurations. It found substantial variability across people, settings, and cultures. For example, a smile can have social meanings beyond happiness, while fear and anger can appear with different facial patterns in different situations. Such variability applies with particular force to isolated, extremely brief movements, where surrounding cues are limited. For a fuller treatment of cultural evidence, see Are Facial Expressions Universal?.
Emotion and facial movement are different measurements
A person's report of fear is evidence of an experienced feeling. Facial action coding is a measurement of visible behavior. Heart rate, skin conductance, and breathing reflect aspects of physiology. Each measure can contribute to a larger study, but none alone offers a complete or unique measure of emotion. Researchers can test associations between these variables; asserting that one caused another or reveals the entire emotional state requires a different design. A short expression also cannot be translated into an exact level of emotional intensity simply because its movement is observable.
Can someone show one emotion while feeling another?
Yes. Emotional displays are shaped by social expectations, interaction goals, and deliberate or practiced control. Someone may smile when embarrassed, stay calm while frightened, or express sympathy without experiencing the same feeling as another person. A briefly visible facial change could be compatible with several interpretations, especially without context. Even if an observer correctly codes a movement commonly associated with fear, the finding by itself does not establish whether the person felt fear, hid fear, recalled fear, or communicated something socially meaningful.
Can Microexpressions Detect Lies? What the Experiments Show
The most consequential popular claim is that a fleeting facial movement betrays a liar. The scientific record is mixed and highly dependent on study design. Some experiments identified associations between particular short-duration negative expressions and deceptive behavior; others found very rare microexpressions, weak observer accuracy, or training effects that failed to improve lie detection. Evidence for a laboratory association and evidence for a reliable decision rule in interviews, hiring, security, therapy, or court proceedings are different standards.
Porter and ten Brinke: brief expressions were uncommon
In a 2008 experimental study, Porter and ten Brinke analyzed 104,550 video frames from 697 facial expressions produced while people responded to emotional photographs and were instructed to be genuine, simulate, neutralize, or mask expressions. Under their narrow microexpression definition, the events occurred in approximately 2% of expressions and were displayed by about 21.95% of participants. Emotional inconsistencies could occur, but many lasted longer than the traditional microexpression interval. Untrained viewers performed only slightly above chance at identifying deception. This is evidence for the possibility of partial emotional leakage under the particular experimental manipulations, not evidence of a frequent or universal lie tell.
Matsumoto and Hwang: a more favorable but narrow finding
A 2018 exploratory mock-crime study found that short negative expressions lasting up to 400 or 500 milliseconds differed statistically between people who intended a deceptive act and truth tellers. The much narrower intervals up to 200 or 300 milliseconds did not show the same discrimination. The study therefore matters for how researchers define the duration of an event. It also shows why headlines stating simply that microexpressions detect lies erase necessary detail. The sample and specific checkpoint-like scenario limit how confidently results generalize to everyday conversation or high-stakes individual judgments.
The authors analyzed a specific prospective-intent deception task rather than ordinary stories told across many settings. The published paper itself identified limited sample size as a concern for testing demographic moderation and replication. The authors also had affiliations with a commercial enterprise related to emotion recognition training, which readers should consider when evaluating applied implications; an affiliation alone does not invalidate the study. The appropriate conclusion is that some duration-defined facial measures showed promising discrimination in this experimental setting and require independent replication. Read the primary study.
The broader deception literature
A large meta-analysis by Bond and DePaulo synthesized 206 documents with 24,483 judges and estimated average unaided lie–truth judgments at approximately 54% correct across the examined studies. That estimate applies to a broad body of deception judgments, rather than specifically to microexpression readers. It provides important background: confident beliefs about lie-reading ability often exceed observed performance. Reviews of nonverbal deception cues find that many cues are faint, inconsistent, or affected by situational factors. Vrij and colleagues (2019) conclude that relying on observed body language alone is a weak approach to deception detection.
A useful causal point follows. Fear, discomfort, or embarrassment can occur while a person speaks truthfully, just as deliberate deception can occur with little outward tension. Thus even a valid signal of fear would not, by itself, be a valid signal of lying. It is a category error to move from a facial observation to a feeling, and from a feeling to guilt, without testing both inferential steps.
Why occasional group differences are insufficient for individual lie judgments
Suppose a study reports that a certain movement is more common in a group instructed to deceive. The two groups may still overlap substantially. Many truthful speakers may display the movement, while many deceptive speakers may not. A useful diagnostic tool must perform well on sensitivity, specificity, calibration, and positive predictive value in representative settings. Real screening environments may have low base rates of deception; even moderately informative signals can then produce many false accusations. An investigator cannot infer a person's honesty from a single momentary pattern just because an average difference was reported.
Does Microexpression Training Work?
There is a meaningful distinction between learning to classify images in a training task and improving the interpretation of people in everyday life. Practice can increase speed and accuracy on similar stimuli, particularly when the learner sees clearly posed, labeled facial configurations. This is a measurable skill, but its usefulness depends on whether it transfers to spontaneous expressions and more complex settings.
Improvement on trained recognition tasks
In Shen, Wu, and Fu (2012), participants improved recognition of rapidly presented expression images, and duration strongly influenced correct identification. More recently, a randomized study by Döllinger and colleagues (2021) assigned 67 participants to multimodal emotion training, facial microexpression training, or an active control. Each targeted program improved the specific kind of recognition on which participants practiced. Crucially, the study found no transfer to its separate measure involving emotional cues from patients. Thus a gain on a test of rapidly displayed faces is real within that task but cannot automatically be translated into professional interpersonal expertise.
Recognizing faces is not the same as detecting deception
In Curtis's 2021 study, 55 participants, including counseling psychology graduate students and undergraduates, were randomized to facial microexpression training or a control condition. Training improved some emotion-recognition outcomes, especially for contempt, anger, and fear, but did not produce a significant improvement in deception detection. The study illustrates the gap between recognizing labeled facial patterns and assessing truthfulness. One modest experiment cannot prove that every possible training program fails; it does directly challenge claims that this particular type of training establishes a reliable lie-detection skill.
A broader meta-analysis of deception-detection training found a small-to-medium overall effect across varied instructional approaches, with larger benefits from training focused on verbal content cues than from nonverbal or related cue training. That result should not be described as proof for microexpression programs: the meta-analysis combines different kinds of instruction and outcome measures. Transfer, durability, independent validation, and potential harm from false confidence deserve attention whenever such courses are offered for professional use.
What would meaningful training validation look like?
A training study should compare a genuine control group, randomly assign participants when feasible, use unfamiliar people and naturally occurring interactions during testing, blind assessors, and follow trainees over time. It should measure performance on tasks that matter outside the training environment and track both false accusations and overlooked instances. Evaluation should report who benefits, under what conditions, and by how much relative to ordinary conversation or better validated assessment methods. Scores on a proprietary recognition quiz are insufficient as a stand-alone professional credential.
Can AI or a Camera Reliably Identify Microexpressions and Emotions?
Computer vision systems can detect and classify some facial action patterns and changes across video frames. High-speed cameras and algorithms make it easier to locate events too brief for unaided observers to notice. Yet automated detection has at least three separate questions: Did a facial movement occur? Does a model's assigned expression label match a reference annotation? Does that movement establish the person's subjective emotion or honesty? Performance on the first or second does not validate the third.
Many image-recognition benchmarks use curated datasets, posed configurations, labeled clips, or constrained lighting and camera perspectives. The labels may represent what annotators inferred from behavior, not an independently observed inner feeling. Generalization to natural conversations, diverse populations, facial coverings, low-quality footage, and changing social contexts needs direct testing. The 2019 scientific review of facial emotion inference explains why treating facial configurations as transparent readings of emotion is conceptually and empirically problematic. Specialized evaluation of AI emotion recognition belongs to a separate EN-23 article.
AI expression detection versus subjective experience
A detector can report an estimated action unit, an annotated expression category, or a probability under a statistical model. It cannot establish that a person subjectively experienced the category, intended to deceive, or secretly held a particular belief solely from the video. Even a high classifier accuracy score may be inflated by similarity between training and testing individuals, dataset artifacts, selection of unusually expressive examples, or an unrealistic distribution of rare events. Ethical claims about workplace, educational, clinical, or security deployment therefore require external validity and prospective harm assessments, not only technical benchmark results.
Why Interpreting a Fleeting Facial Cue Is Difficult
Context changes what a movement means
The same smile or tightened brow can serve different functions in a family conversation, professional meeting, medical appointment, or stressful interview. What was said, what occurred immediately before, relational expectations, social roles, and the person's baseline behavior all contribute information. Facial actions should be examined in combination with timing, speech, circumstances, and the person's own account where appropriate. A context-free freeze frame removes exactly the information needed to evaluate competing explanations. See our overview of emotion recognition across faces, voices, and context.
Cultural and individual differences
Research on facial communication supports some recurrent patterns while also documenting substantial variability in production, interpretation, and display rules. People differ in how expressively they use their faces, how they manage social attention, and how they respond to particular emotional situations. A culturally familiar facial cue may be overinterpreted when the observer lacks experience with the speaker's communicative environment. The full theoretical debate extends beyond microexpressions; see Basic Emotion Theory and the cross-cultural evidence review.
Disability, neurodiversity, and health considerations
Facial mobility and expressiveness may vary for many reasons, including motor function, fatigue, medication effects, neurodevelopmental differences, pain, and individual communicative style. These differences should not be used to infer intent or characterize credibility. Similarly, difficulty naming one's internal feelings should never be deduced from someone else's interpretation of a facial movement. Our article on alexithymia explains the research construct of identifying and describing feelings in its appropriate psychological context.
Expectations and confirmation bias
Once an observer suspects deception, a brief facial movement may seem like confirmation, even though the same movement would have passed unnoticed in an ordinary conversation. Selectively replaying the one suspicious frame from a long interview creates a serious multiple-comparisons problem: many brief movements become opportunities to find an apparently incriminating cue. A scientifically responsible analysis specifies the coding and decision rules in advance and tests whether the cue predicts outcomes in new people and settings.
Practical Implications: How to Respond to Brief Facial Changes
For everyday interaction, a brief facial change is best treated as a possible prompt for curiosity rather than a verdict. If someone seems uncomfortable, a respectful question such as 'Would you like to talk about that?' can obtain information the facial movement itself does not provide. In professional or clinical conversations, attend to what the person says, the setting, and their expressed preferences. Someone may prefer privacy or may have ordinary reasons not to explain every expression.
For practitioners, investigators, or organizations, isolated microexpressions should not be used as a sufficient basis for clinical conclusions, credibility determinations, accusations, screening exclusions, or disciplinary action. Valid assessment requires methods suited to the actual question and an explicit appraisal of errors. The research base supports careful observation, transparent uncertainty, and corroboration; it does not establish a universal facial shortcut to emotions or lies.
Common Misconceptions About Microexpressions
Myth: Everyone briefly reveals every hidden emotion
No study establishes that each concealed emotional episode produces an observable, recognizable microexpression. The frequency depends on measurement definitions and experimental situations. The Porter and ten Brinke study found microexpressions in only a small percentage of analyzed expressions under its narrow definition.
Myth: A short expression always reveals the person's real feeling
Duration identifies a temporal feature. Emotion interpretation requires contextual and independent evidence. Multiple emotional and nonemotional processes can be associated with similar muscle movements. The comprehensive review cautions against one-to-one inference from particular facial configurations.
Myth: Training turns observers into reliable lie detectors
Recognition practice can improve scores on recognition tasks. That improvement is not equivalent to reliable detection of dishonest statements in real life. Curtis found recognition gains without significant deception-detection improvement, and Döllinger and colleagues found task-specific gains without the tested transfer effect.
Myth: Video or AI can reveal the truth that people conceal
Algorithms and careful frame analysis can help detect patterns of movement, but the inference from movement to emotion and from emotion to veracity remains unproven as a general high-accuracy procedure. More precise observation does not eliminate ambiguity in what is being observed.
What the Evidence Supports—and What It Leaves Open
There is established evidence that facial movements can be objectively studied at fine timescales; that people can be trained to recognize certain rapid, labeled expression stimuli; and that observable facial behavior has social and emotional relevance. There is limited, context-dependent evidence that some categories of short-duration facial activity differ between truthful and deceptive groups in particular laboratory tasks. There is insufficient evidence for a dependable, general-purpose method that reads a person's true emotion or detects a lie from microexpressions alone. These judgments concern distinct research questions and should not be collapsed into one conclusion.
Important open questions include how often naturally occurring microexpressions arise in everyday social interaction; whether findings generalize across populations and tasks; which duration and intensity rules yield replicable measurements; how facial signals interact with speech and context; and whether training creates skills beyond the stimuli used during instruction. Progress depends on independent replication, preregistered protocols, validated real-world outcomes, and careful assessment of the consequences of mistakes. The wider science of emotion begins with the definition, components, and functions of emotions.
Frequently Asked Questions
Are microexpressions real?
Very brief facial movements are observable and can be studied in video records. The scientific debate concerns how to define them, how often they occur, and how confidently they indicate a specific emotion or deceptive intention. Their existence as movements should not be confused with the stronger claim that they reliably reveal psychological truths.
How long does a microexpression last?
The narrow traditional interval is about 40–200 milliseconds, but researchers have also tested upper limits of 300, 400, and 500 milliseconds. Always check the definition used by the specific study. A difference in cutoff can substantially change the apparent frequency and predictive findings. Matsumoto and Hwang (2018) systematically compared several thresholds.
Can you spot a microexpression with the naked eye?
Some brief movements may be noticed, particularly with practice or repeated observation, but reliable frame-level coding often depends on video review. Unassisted observers may miss rapid events, and noticing an event remains separate from knowing why it occurred.
Do microexpressions show hidden anger or fear?
A movement can be compatible with an emotion-related interpretation, but it does not uniquely establish anger or fear. Expression, feeling, bodily change, and social context vary from person to person. The person's own report and relevant circumstances provide essential additional information.
Can microexpressions tell if somebody is lying?
The research has not established microexpression observation as a reliable stand-alone lie detector. A few task-specific experiments reported associations, but broader deception research and training evaluations do not support confident case-by-case decisions from fleeting facial cues. The major review of deception research explains the limitations.
Are microexpressions involuntary?
Some facial changes occur with limited voluntary control, and emotional leakage is a plausible mechanism for certain events. Researchers cannot determine from an isolated movement alone whether the person consciously controlled it, what they felt, or whether they intended to conceal an emotion.
What is the difference between a microexpression and a subtle expression?
A microexpression is defined mainly by short duration. A subtle expression is defined mainly by low intensity or incomplete movement. An expression can be both brief and subtle, brief and strong, or relatively long and subtle. The two labels answer different measurement questions.
Do people who are better at recognizing microexpressions detect lies better?
Not necessarily. Research shows that improvement in emotion-expression recognition may fail to transfer to deception detection. Recognizing a labeled face is one task; determining whether a claim is true is another. Curtis (2021) examined this distinction experimentally.
Can microexpression analysis diagnose a mental health condition?
No brief facial cue is, by itself, a clinical diagnosis. Assessing symptoms and disorders requires appropriate history, validated criteria, and qualified clinical judgment. Facial behavior can be studied as one type of information, but no one-to-one interpretation should be applied to individuals.
Can AI emotion recognition interpret microexpressions?
Software can be trained to detect or classify some rapid facial patterns under defined conditions. That capacity does not establish direct access to subjective emotion, hidden intent, or truthfulness. The output is a model-based inference conditioned on data, labeling rules, and context.
What is the safest way to interpret someone's brief expression?
Notice it without turning it into a conclusion. Consider the surrounding conversation, avoid stereotypes, and ask respectful questions when appropriate. For consequential decisions, use corroborating information and methods validated for that purpose.
Related Articles
Emotion Recognition: Faces, Voices, and Context explores inference from multiple human social signals and the limits of reading emotion from a single cue.
Are Facial Expressions Universal? Cross-Cultural Evidence examines variability and shared patterns across cultures, while Basic Emotion Theory: Six Emotions and Scientific Debate explains a major theoretical framework behind facial-expression research.
What Are Emotions? Psychological Definition, Components, and Functions establishes the broader construct; Alexithymia: Difficulty Identifying and Describing Feelings examines a separate individual-differences topic.
References
Barrett, L. F., Adolphs, R., Marsella, S., Martinez, A. M., & Pollak, S. D. (2019). Emotional expressions reconsidered: Challenges to inferring emotion from human facial movements. Psychological Science in the Public Interest, 20(1), 1–68. https://doi.org/10.1177/1529100619832930
Bond, C. F., Jr., & DePaulo, B. M. (2006). Accuracy of deception judgments. Personality and Social Psychology Review, 10(3), 214–234. https://doi.org/10.1207/s15327957pspr1003_2
Curtis, D. A. (2021). Deception detection and emotion recognition: Investigating F.A.C.E. software. Psychotherapy Research, 31(6), 802–816. https://doi.org/10.1080/10503307.2020.1836424
Döllinger, L., Laukka, P., Högman, L. B., Bänziger, T., Makower, I., Fischer, H., & Hau, S. (2021). Training emotion recognition accuracy: Results for multimodal expressions and facial micro expressions. Frontiers in Psychology, 12, 708867. https://doi.org/10.3389/fpsyg.2021.708867
Hauch, V., Sporer, S. L., Michael, S. W., & Meissner, C. A. (2016). Does training improve the detection of deception? A meta-analysis. Communication Research, 43(3), 283–343. https://doi.org/10.1177/0093650214534974
Matsumoto, D., & Hwang, H. C. (2018). Microexpressions differentiate truths from lies about future malicious intent. Frontiers in Psychology, 9, 2545. https://doi.org/10.3389/fpsyg.2018.02545
Porter, S., & ten Brinke, L. (2008). Reading between the lies: Identifying concealed and falsified emotions in universal facial expressions. Psychological Science, 19(5), 508–514. https://doi.org/10.1111/j.1467-9280.2008.02116.x
Shen, X.-B., Wu, Q., & Fu, X.-L. (2012). Effects of the duration of expressions on the recognition of microexpressions. Journal of Zhejiang University-SCIENCE B, 13(3), 221–230. https://doi.org/10.1631/jzus.B1100063
Vrij, A., Hartwig, M., & Granhag, P. A. (2019). Reading lies: Nonverbal communication and deception. Annual Review of Psychology, 70, 295–317. https://doi.org/10.1146/annurev-psych-010418-103135
Zloteanu, M., Bull, P., & Krumhuber, E. G. (2021). Veracity judgement, not accuracy: Reconsidering the role of facial expressions, empathy, and emotion recognition training on deception detection. Quarterly Journal of Experimental Psychology, 74(5), 910–927. https://doi.org/10.1177/1747021820978851