New psychological research suggests that negative facial expressions do more than just change the mood of an interaction – they can actually blur our ability to see who is causing what in social situations, with striking implications for everything from arguments at home to reading emotions at work.
How faces quietly shape our sense of cause and effect
Everyday social life runs on quick predictions. You smile at a colleague, you expect a smile back. You raise your voice, you fear a defensive reply. Psychologists call this skill “contingency learning”: judging how likely a certain outcome is, given a particular signal or cue.
A new series of experiments, published in the Quarterly Journal of Experimental Psychology, suggests that our ability to judge these links is not neutral. It appears to be disrupted when the faces we see look sad, angry or afraid.
Across four experiments, people consistently judged cause-and-effect links as weaker when negative facial expressions were involved.
The research team, led by Rahmi Saylik at Mus Alparslan University with colleagues from Brunel University London and the University of Oxford, set out a simple question: does the emotional tone of a face – its “valence” – make us better or worse at detecting causal relationships between people?
The experiment: rapid faces, hidden patterns
To test this, the researchers built a computer-based task that looks almost like a pared-down social media feed: a rapid stream of images, flashing by one after another.
In the emotional conditions, participants saw two faces on screen. One acted as the “sender”, the other as the “receiver”. The key question was whether the expression on the first face appeared to bring about the expression on the second.
Over a series of short blocks, the team secretly controlled the statistical structure of the stream. Sometimes, the first expression strongly predicted the second. In other blocks, there was little or no genuine connection between the two.
Experiment 1: happy faces, sad faces and simple shapes
In the first experiment, 107 volunteers watched streams involving:
➡️ The incredible story of Craighead’s underground lake found by a child
➡️ Psychology reveals why emotional balance can feel fragile even in stable periods
➡️ People Who Are Often Bored Are Missing This Surprising Skill
➡️ Gardeners who vary planting depth slightly see more balanced growth
➡️ After 60, is it better to wake up early or sleep longer ?
- happy faces
- sad faces
- geometric shapes (used as a neutral, non-social control)
After each block, participants rated how strong they believed the causal link was, using a scale going from negative to positive.
People could tell the difference between strong and weak links, but sad faces consistently dragged their ratings down.
When faces looked sad, participants rated the causal connections as weaker than when they saw happy faces or basic shapes. Their ratings also lined up less accurately with the actual statistics the researchers had built into the task.
This suggested that sadness did not just change the mood of the trial; it actively muddied people’s sense of cause and effect.
Experiment 2: ruling out simple visual differences
One obvious criticism is that faces and shapes do not look alike. Faces are complex, expressive, full of fine detail; shapes are simple and abstract. Perhaps sad faces simply made the task visually harder.
To address this, the team redesigned the images for 82 new participants:
- all images were black-and-white
- faces were presented through oval cut-outs, to match the shape of the non-face stimuli
- patterned shapes were created to mimic the presence or absence of a “feature”, like an emotional expression on a face
Even when the visuals were tightly controlled, the key pattern stayed put. Sad faces again produced lower causality ratings than happy faces or the patterned shapes.
Happy faces did not supercharge learning; it was the negative expressions that consistently interfered.
There was no reliable difference between happy faces and the neutral, patterned shapes. This pushed the interpretation away from “happy faces help” and squarely towards “negative faces hinder”.
Are people really doing the maths, or just counting pairs?
Another challenge came from a long-standing idea in psychology known as the “pairing hypothesis”. This view suggests that when judging causality, people often rely on something basic: how often two things appear together, rather than on deeper statistical reasoning about probabilities.
To test this, the researchers ran a third experiment with 90 participants. This time, they unlinked the number of pairings from the true predictive strength of the cue.
| Condition type | Number of pairings | Actual predictive strength |
|---|---|---|
| High-pairing, weak link | Many cue–outcome pairs | Low causality |
| Low-pairing, strong link | Few cue–outcome pairs | High causality |
If people were just counting co-occurrences, they should rate the high-pairing condition as more causal, regardless of the underlying statistics.
Instead, participants tracked the true contingency. They gave higher ratings when the statistical relationship was strong, even if the events appeared together less frequently.
People were sensitive to the underlying probabilities – and negative faces still dulled their judgments.
Sad expressions continued to reduce perceived causal strength, despite the more demanding statistical structure of the task.
Beyond sadness: anger, fear and a broader pattern
The final study widened the lens. Was this effect unique to sadness, or part of a broader problem with negative emotional signals?
In Experiment 4, 51 participants completed similar tasks, but this time with faces expressing happiness, anger and fear.
Negative emotions again disrupted causal learning. When participants looked at angry or fearful faces in situations where a positive causal link genuinely existed, they gave lower ratings of causality than when the faces were happy.
Angry and fearful expressions, like sad ones, consistently weakened people’s sense of “A leads to B”.
The data pointed towards a general influence of negative valence on contingency learning, rather than an effect driven by sadness alone.
Why negative faces drain mental bandwidth
The researchers interpret these results in terms of attention and limited mental resources. Threatening or distressed faces are known to grab attention rapidly. From an evolutionary standpoint, that makes sense: noticing anger, sorrow or fear can signal danger or social breakdown.
But that attention comes at a cost. A negative face may trigger thoughts such as “Is this about me?”, “Am I safe?”, or “What went wrong here?”. It may also produce physical arousal: faster heart rate, muscle tension, a sense of unease.
When we react internally to someone’s distress or anger, fewer resources are left for tracking who caused what.
In the lab task, that meant participants noticed the faces, but had less capacity to compute the exact relationship between cue and outcome. Happy faces, by contrast, tend to signal safety and reward. They draw attention without kicking off as much inner turmoil, leaving more bandwidth for quiet statistical calculations.
What this means in everyday life
The findings raise awkward questions about everyday misunderstandings. During an argument, when faces are angry or distressed, the people involved may be genuinely less able to follow the chain of cause and effect.
- A manager frowning when giving feedback may leave staff less able to see how their actions influenced the outcome.
- Parents arguing in front of a child may struggle to track who actually escalated the conflict.
- Partners in a tense conversation might fail to notice small conciliatory gestures because their brains are overloaded by negative expressions.
This does not excuse behaviour, but it does hint at a hidden cognitive blind spot: when emotions turn dark, our perception of “who did what to whom” quietly degrades.
Limitations and open questions
The work relied on static, computer-presented images and mostly neurotypical university students. Real-life interactions are more chaotic: faces move, voices rise and fall, context shifts second by second. Non-emotional shapes, no matter how carefully designed, can never match the rich meaning of a human face.
Future experiments could measure response times to see whether negative faces slow down causal judgments, and use physiological tools such as heart-rate monitoring or skin conductance to track arousal during the task. Researchers are also keen to see how these effects play out in people with anxiety, depression or autism, who often report difficulties reading social situations.
Key concepts behind the research
A few technical terms sit quietly underneath these results:
- Valence: the emotional “direction” of a signal – broadly positive (pleasant, safe) or negative (threatening, unpleasant).
- Contingency: the statistical relationship between a cue and an outcome, often expressed as “How much more likely is B when A happens?”.
- Contingency learning: the process of gradually inferring these probabilities from repeated experiences.
A simple way to picture the task is as a social weather forecast. Participants were, in effect, trying to say, “When this face appears, how likely is it that the next face will show this emotion?” Negative expressions seemed to throw static onto that forecast, leaving people less confident and less accurate in their predictions.
For anyone involved in emotionally charged communication—teachers, therapists, managers, parents—this research suggests a subtle risk: the more negative the expressions on display, the harder it may be for everyone involved to clearly see the causal patterns that could help resolve the situation.
Originally posted 2026-02-05 16:45:21.