Is the Face a Window to the Soul? Investigation of the Accuracy of Intuitive Judgments of the Trustworthiness of Human Faces

Porter et al. (2008) examined whether people can accurately judge the trustworthiness of strangers from facial appearance alone, comparing Nobel Peace Prize recipients and Order of Canada humanitarians against America's Most Wanted criminals. Participants viewed 34 faces under ultra-brief (100 ms) or extended (30 s) exposure and rated trustworthiness, then classified each face into a target group after being informed of the categories. The study found that accuracy was essentially at chance for identifying criminal (untrustworthy) faces and only modestly above chance for humanitarian (trustworthy) faces, with exposure duration having negligible effect.

Laboratory analog study in which participants viewed 34 photographs of Nobel Peace Prize recipients/humanitarians and America's Most Wanted criminals for either 100 ms or 30 s, rated perceived trustworthiness, and then classified each face by group after category disclosure.

Porter's 2008 study tested the ancient assumption that trustworthiness can be accurately assessed from facial appearance by using extreme groups - Nobel Peace Prize winners/Order of Canada humanitarians versus America's Most Wanted criminals. Participants viewed faces under brief (100ms) or extended (30s) exposure conditions and made trustworthiness ratings followed by explicit group classifications. The two-phase design elegantly separated intuitive judgments from deliberative processing while testing whether thin-slice judgments differ from longer evaluations. Results indicated that participants based their decisions on perceived kindness and aggressiveness, though the study challenged assumptions about the accuracy of face-based character assessments.

This research has direct relevance for polygraph practice by questioning the reliability of examiner impressions based on subject appearance and highlighting potential biases in credibility assessment that could affect examination outcomes.

Background & Context

The capacity to assess another person's character from their face alone is one of the most ancient and deeply held assumptions in human social cognition. Long before formal deception-detection tools such as the polygraph were developed, people relied on facial appearance to make rapid judgments about who could be trusted. Porter and colleagues' 2008 study set out to rigorously test whether these intuitive assessments bear any relationship to actual trustworthiness — a question with profound implications for criminal justice, credibility assessment, and the science of deception detection.

Although trustworthiness judgments based on a stranger's face occur rapidly, their accuracy had previously been unknown. Prior research had established the speed and consistency of such judgments but largely neglected their validity. Inferences of trustworthiness, based on a stranger's face, occur within 38 ms, and while confidence in judgment accuracy increases with more lengthy exposure to a face, assessments remain virtually unchanged. This paradox — rapid, confident judgments that may nonetheless be wrong — sits at the heart of Porter et al.'s investigation.

The study arrives at a moment of growing interdisciplinary interest in thin-slice judgments and social perception. Researchers had shown increasing interest in assessments of trustworthiness, devoting much attention to whether trustworthiness can be detected from a person's facial appearance, though the results had been investigated along diverse behavioral dimensions with great inconsistency. By targeting the extreme ends of the trustworthiness spectrum — the most celebrated humanitarians and the most notorious criminals in society — the authors created an unusually powerful test of the intuitive accuracy hypothesis.

Research Design & Methodology

The researchers examined the accuracy of trustworthiness judgments of the faces of two groups differing in trustworthiness: Nobel Peace Prize recipients and Order of Canada humanitarians versus America's Most Wanted criminals. Participants viewed 34 faces each for either 100 ms or 30 seconds and rated their trustworthiness. Subsequently, participants were informed about the nature of the two groups and estimated group membership for each face.

This two-phase design was methodologically elegant. The first phase captured intuitive rating of perceived trustworthiness on a scale, while the second phase — after disclosure of the two categories — tested whether participants could correctly classify each face into the criminal or humanitarian group. The manipulation of exposure duration (100 ms vs. 30 seconds) directly tested the "thin-slice" hypothesis: whether brief, automatic processing matches or diverges from longer, more deliberative processing.

Key design elements include:

• Target stimuli: 34 photographic faces drawn equally from Nobel Peace Prize winners/Order of Canada humanitarians and America's Most Wanted criminals
• Exposure conditions: Ultra-brief (100 milliseconds) versus extended (30 seconds) viewing times, assigned between subjects
• Two-stage procedure: First, trustworthiness ratings; second, explicit group classification after being informed of the two target categories
• Institutional base: Conducted at Dalhousie University, Department of Psychology, Halifax, Nova Scotia, Canada
• DOI: 10.1037/0008-400X.40.3.171, published in Canadian Journal of Behavioural Science, Vol. 40, No. 3, pp. 171–177

Participants were found to rely on perceived kindness and aggressiveness to inform their intuitive decisions. The use of real-world, publicly known figures at the extremes of trustworthiness served as a naturalistic ground-truth criterion, bypassing the need for laboratory-constructed deception scenarios.

Results & Key Findings

The headline finding is striking in its implications: participants' intuitive judgments of trustworthiness were only marginally better than random chance, and significantly asymmetric across target type. Judgments formed with extremely brief exposure were similar in accuracy and confidence to those formed after a long exposure. However, initial judgments of untrustworthy (criminals') faces were less accurate (M = 48.8%) than were those of trustworthy faces (M = 62.7%). Judgment accuracy was above chance for trustworthy targets only at Time 1, and only slightly above chance for both target types at Time 2.

Key numerical outcomes from the study:

• Criminal face accuracy (100 ms exposure): M = 48.8% — essentially at or below the 50% chance baseline
• Humanitarian face accuracy (100 ms exposure): M = 62.7% — modestly above chance, but far from reliable
• Effect of extended viewing (30 s): Both target types rose only slightly above chance; the longer exposure conferred negligible benefit
• Exposure duration effect: No meaningful difference in accuracy or confidence between 100 ms and 30-second conditions
• Overall pattern: Accuracy ranges around chance level, particularly for untrustworthy targets

Research showed that faces of criminals were judged as less trustworthy than faces of exemplary citizens, but only to a small degree and subject to contextual restrictions. The asymmetry is theoretically important: people show a modest ability to recognize benevolent faces, but are essentially unable to identify criminal faces from appearance alone. These ratings should not be considered a valid source of information, as judgments range around chance level.

Discussion & Significance

The Porter et al. (2008) findings deliver a direct empirical challenge to one of the most persistent lay beliefs in human cognition: that the face is a reliable "window to the soul." Despite the fact that participants were judging faces at the extreme ends of the trustworthiness continuum — Nobel Peace Prize laureates versus America's Most Wanted — their accuracy was barely distinguishable from flipping a coin when identifying untrustworthy individuals. This is arguably the most favorable possible test scenario for the intuition hypothesis; real-world judgments involving more ambiguous targets would presumably fare even worse.

The equivalence of accuracy across exposure durations carries a particularly important implication for forensic contexts. Inferences of trustworthiness based on a stranger's face occur within 38 ms and remain virtually unchanged with more lengthy exposure; the instantaneous and enduring nature of these assessments can color subsequent decision-making. In other words, once a first impression forms, it is highly resistant to revision — even when observers are given thirty times more viewing time. This "anchoring" effect of initial trustworthiness impressions poses a serious risk in legal and investigative settings.

The downstream consequences of these flawed but persistent judgments have been documented by Porter and colleagues in related work. Mock jurors needed significantly less evidence to reach a guilty verdict when the face of a defendant appeared to be more untrustworthy in severe crimes such as murder cases. Even during judicial activities, perceptions of facial trustworthiness could be used to predict decisions about whether to sentence a convicted murderer to death or life imprisonment. The 2008 study thus anchors a broader program of research demonstrating that inaccurate facial trustworthiness judgments have life-altering consequences in justice systems.

Limitations & Considerations

Several methodological considerations affect how broadly these results can be interpreted:

• Extreme category design: Using Nobel Peace Prize recipients and America's Most Wanted criminals represents the far poles of trustworthiness. Real-world credibility assessment rarely involves such clear-cut categories, and accuracy may be even lower with more typical targets.
• Static photographs: The study used still images, excluding dynamic facial cues (microexpressions, movement, vocal tone) that occur in genuine person-to-person interaction.
• Confounded variables: Photographs of famous individuals may carry contextual cues (e.g., photographic quality, media presentation style, racial composition, age) that could influence judgments independently of facial trustworthiness per se.
• Participant sample: As with most psychological studies of the era, participants were likely drawn from university populations, which limits demographic generalizability.
• Ground truth definition: "Trustworthy" and "untrustworthy" are operationalized through public recognition (awards vs. criminal listing) rather than behaviorally validated measures of honesty, which is a necessary but imperfect proxy.

Although it is likely that everyone's honesty fluctuates because lying is such an important part of social interaction, the researchers were interested in examining judgments of broader character. This framing is important: the study measures dispositional impressions of trustworthiness, not the detection of a specific deceptive act — a distinction that should be kept in mind when extrapolating to polygraph or real-time deception contexts.

Practical Applications

For polygraph examiners and credibility assessment professionals, this study delivers a critical cautionary message: visual impressions of a subject's trustworthiness formed before or during an examination should not be treated as meaningful data. The research demonstrates that even when observers are given extended time to examine a face, their accuracy at identifying untrustworthy individuals hovers around the 50% chance baseline. An examiner who "reads" a subject as deceptive based on facial appearance is engaging in a cognitively compelling but empirically unsupported inference. Structured, instrument-based assessment is essential precisely because human intuition fails in this domain.

For consumers of credibility assessment services — including those involved in legal proceedings, hiring decisions, or security screening — these findings underscore the danger of relying on appearance-based credibility cues. Porter, Gustaw, and ten Brinke (2010) showed that mock juries required less evidence to arrive at a guilty verdict when a defendant looked untrustworthy, and Wilson and Rule showed that convicted criminals who looked untrustworthy were more likely to have received a death rather than a life sentence. The 2008 Porter et al. study is the empirical foundation for understanding why these biases are so dangerous: they feel valid but are demonstrably unreliable. Only validated, criterion-referenced assessment tools can compensate for the systematic shortcomings of human facial trustworthiness judgment.