English (en)
If We Cannot Read Minds, How Well Can We Predict What Someone Will Do Next from Their Behavior?
5-second answer: You cannot see inside another person’s head. But reply timing, scheduling style, whether they offer alternatives, and how late they wait before deciding are observable habits. Those cues can update the probability of what they do next. This is less mind reading and more reading an operating log.
30-second overview: Someone said, “I’ll contact you when I know my schedule.” Asked when they would know, they answered, “I don’t know.” From that tiny exchange came a prediction: “This person may suddenly message the night before saying they are free tomorrow.” Later, that is almost exactly what happened. Psychology offers several useful frameworks for why. Research on thin-slice judgments shows that short observations can contain predictive information; a classic meta-analysis reported an overall effect size of r=.39 across 38 outcomes. That does not mean 39% accuracy. Past behavior can also predict future behavior, especially in recurring contexts. And experience becomes trustworthy intuition only when the environment contains learnable regularities and provides feedback. What was predicted here was not a soul, motive, or secret feeling. It was how a person tends to run their schedule.
1. “I’ll Let You Know” → “When?” → “I Don’t Know”
Three messages are not much data.
They do not reveal personality, private motives, or life circumstances.
But they do reveal one operational fact:
there is no declared point at which the schedule is expected to become fixed.
That supports a modest prediction: maybe this person does not lock plans days ahead, but contacts others when availability suddenly becomes clear.
Then, later, a message arrives the night before: “I can do tomorrow morning.”
Not prophecy.
The schedule leaked the scheduling style.
2. The Target Was Operations, Not the Mind
“Does this person secretly dislike me?” “Are they angry?” “What are they really thinking?”
Those are hard because mental states are not directly observable.
But these are observable:
- how quickly they reply
- how far ahead they confirm plans
- whether they offer alternatives
- when they update after saying “I don’t know”
- how they communicate changes
Predicting repeated procedures is much more realistic than claiming access to hidden feelings.
Do not read the soul. Read the workflow.
3. Thin-Slice Judgment: Even Seconds Can Contain Information
The thin-slice literature asks whether very brief observations of behavior can support useful judgments.
Ambady and Rosenthal’s 1992 meta-analysis examined observations shorter than five minutes and their relation to later objective outcomes.
Across 38 outcomes, the overall effect size was r=.39.
Interestingly, observations shorter than 30 seconds were not clearly less accurate than observations lasting four or five minutes.
So a short sample is not necessarily information-free.
But statistics will object loudly if we translate this into “people are right 39% of the time.”
4. r=.39 Is Not 39% Correct
r=.39 is a correlation, not a hit rate.
It means judgments based on brief observations and real outcomes were moderately related across those studies.
Different studies predicted different things and used different types of outcomes.
Therefore there is no universal answer to:
“What percentage of someone’s next behavior can humans predict?”
Accuracy depends on the target, context, base rates, cues, and experience.
Scheduling habits may be fairly predictable while romantic feelings are not.
5. The Lens Model: Hidden Things Are Inferred Through Cues
Brunswik’s lens model, and later lens-model approaches to personality judgment, provide a useful map.
The hidden state itself is not directly visible.
What we see are cues produced by it.
A person who often decides plans very late might leave cues such as:
- no expected confirmation date
- no tentative booking
- sudden contact when a slot opens
A judge uses those cues to estimate an outcome.
Accuracy depends on two things: whether the cue genuinely relates to the outcome, and whether the observer uses the cue appropriately.
This is not a sixth sense.
It is cue quality control.
6. Past Behavior Can Inform Future Behavior
Ouellette and Wood’s 1998 review described multiple ways past behavior predicts future behavior.
When contexts repeat and behavior becomes well practiced, previous behavior can reflect habit strength and directly inform what happens next.
Someone who repeatedly schedules at the last minute may do so again in a similar situation.
Someone who consistently offers alternatives when unavailable may do so again.
Exceptions always exist.
But repeated history is better evidence than one dramatic impression.
7. When Does Experience Become Real Intuition?
Experience alone is not enough.
Kahneman and Klein argued that skilled intuition requires at least two important conditions.
- The environment contains regularities that can be learned.
- The person gets enough feedback to learn those regularities.
If you repeatedly handle similar scheduling exchanges and later observe what each person actually does, patterns can become internalized.
If outcomes are never observed, confidence can grow without accuracy.
That is not expertise.
It is just confident guessing with seniority.
8. Repeated Coordination Quietly Becomes Training Data
Dozens or hundreds of similar communications create a natural learning set.
“This wording often leads to a late update.” “This reply pattern usually means nothing will be fixed today.” “This person offers a replacement time immediately.”
No spreadsheet is required for the brain to notice repeated pairings.
In machine-learning language:
message pattern = input
later behavior = label
The human brain builds a tiny classifier while pretending it is just answering messages.
9. What Is Relatively Predictable? Repeated Operations
Behavior prediction works best when the behavior repeats under similar conditions.
Examples include:
- usual reply times
- how early plans are confirmed
- how delays are reported
- whether alternatives are offered
- whether action happens near deadlines
- frequency of schedule changes
- whether explanations are proactive
These are procedural patterns.
They are often more stable than one-off emotional events.
The predictable part is usually not the person’s essence. It is their recurring admin.
10. What Is Hard to Predict? Motives, Feelings, and One-Off Circumstances
A late reply does not uniquely mean dislike.
An uncertain schedule does not prove irresponsibility.
No alternative suggestion does not prove selfishness.
The same visible action can come from many hidden causes: workload, health, family demands, missed notifications, forgetfulness, or simple uncertainty.
This is why predicting the next observable action is often easier than identifying the inner cause of that action.
“Probably a last-minute message” can be reasonable.
“This is what kind of person they are” is a much bigger leap.
11. Remember Only the Hits and Everyone Becomes a Prophet
Confirmation bias is a major trap.
Correct predictions are satisfying, so we remember them.
Wrong predictions are boring, so we quietly delete them from memory.
Soon our internal scoreboard shows a 97% win rate.
To measure real accuracy, record the prediction before the outcome and score every result.
“Will message next week.” “Will wait until the night before.” “Will offer an alternative.”
Then count the misses too.
The natural enemy of prophecy is a spreadsheet.
12. Use Predictions as Probabilities, Not Labels
The safest use of behavioral prediction is operational.
Not:
“This person will definitely contact me late.”
But:
“Late contact seems more likely, so I should not rely on this one option.”
Not:
“This person is unreliable.”
But:
“Their scheduling tends to settle late, so I should state a clear deadline.”
Change the process rather than labeling the identity.
If the prediction is wrong, the cost stays small.
The goal is not to expose a human being.
It is to keep the calendar from exploding.
13. Conclusion: Minds Are Hidden, Habits Leak
The mind is a black box.
We cannot directly read what another person truly thinks.
But black boxes still have inputs and outputs.
People reply, delay, confirm, offer alternatives, and make last-minute changes.
When those patterns repeat, they become evidence about what may happen next.
Research shows that brief observations can carry predictive information. Past behavior can inform future behavior in recurring contexts. And expert intuition needs both regularity and feedback.
So the most accurate summary is:
You cannot read minds or see personality directly. But operational habits leak quite a bit.
The successful prediction was not about the future itself.
It was about how that person tends to decide the future.
Sources
- Ambady, N., & Rosenthal, R. (1992). Thin slices of expressive behavior as predictors of interpersonal consequences: A meta-analysis. Psychological Bulletin, 111(2), 256–274. DOI: 10.1037/0033-2909.111.2.256. https://cir.nii.ac.jp/crid/1361981469152380928?lang=en
- Back, M. D., et al. (2019). Lens and Dual Lens Models. In The Oxford Handbook of Accurate Personality Judgment. https://doi.org/10.1093/oxfordhb/9780190912529.013.4
- Ouellette, J. A., & Wood, W. (1998). Habit and intention in everyday life: The multiple processes by which past behavior predicts future behavior. Psychological Bulletin, 124(1), 54–74. DOI: 10.1037/0033-2909.124.1.54. https://dornsife.usc.edu/wendy-wood/wp-content/uploads/sites/183/2023/10/Ouellette.Wood_.1998_Habit_and_intention_in_everyday_life.pdf
- Kahneman, D., & Klein, G. (2009). Conditions for intuitive expertise: a failure to disagree. American Psychologist, 64(6), 515–526. https://pubmed.ncbi.nlm.nih.gov/19739881/
