Your critical thinking gym
Home
The Commandments
Join the Club
Your critical thinking gym
● Cognitive bias
← From the Latin intendere ("to stretch toward," "to direct the mind"). The philosophical term "intentionality," the mind's "aboutness," was coined by Franz Brentano in 1874, but the specific, testable bias this page covers wasn't isolated and named until psychologist Evelyn Rosset's 2008 study.
The default assumption that someone meant to do what they did: reading an action as deliberate and on-purpose before you've actually ruled out accident, mistake, incompetence, or plain bad luck.
Also called: intention bias, "bias for intentional explanations" (Rosset's own term). Related but distinct: hostile attribution bias, fundamental attribution error, and teleological thinking (Section 09).
The tendency to interpret another person's ambiguous action, one that could plausibly have been either deliberate or accidental, as deliberate by default, and to keep treating it as deliberate unless something forces a second look.
Plain version
Someone bumps into you, cuts you off in traffic, or says something that lands wrong. Before you've gathered a single fact about why, your mind has usually already decided: they meant that. Evelyn Rosset, the psychologist who first isolated this pattern, found that adults carry "an implicit bias to infer intention in all behavior," strong enough that even textbook-accidental actions get read as on-purpose when people describe them off the cuff.
Aliases, and what's actually the same thing
Intention bias and Rosset's own phrase, "bias for intentional explanations," describe the identical pattern; the field hasn't settled on one preferred name, so both appear in the literature.
This isn't the same as three closely related pages
Fundamental attribution error is about explaining someone's behavior by their character rather than their situation, even for behavior nobody thinks was chosen on purpose. Hostile attribution bias, a related research area without its own Filter page yet, is narrower still: it's specifically about reading ambiguous actions as hostile or malicious, not just intentional in general. Teleological thinking doesn't require a specific agent at all, a mountain can be treated as having a "purpose" with nobody having built it, while intentionality bias always needs one particular agent's one particular action. Full distinctions in Section 09.
Scope
Not applicable: the memory-distortion and self-perception scope tags describe other bias families on this site. This one operates specifically on how ambiguous behavior gets read, almost always someone else's behavior, which is why it sits in both the judgment-shortcut and social-perception categories at once.
Gets confused with
Fundamental attribution error, hostile attribution bias, and teleological thinking: precise distinctions in Section 09.
About as automatic as this site's list gets. Reading intent isn't a conclusion you arrive at, it's what the action already looks like the instant you see it, which leaves nothing to catch in the moment.
Why your brain does this
Figuring out what another person or animal wants, right now, is one of the most useful things a social mind can do fast. A default assumption of "on purpose" lets you skip straight to the question that actually matters, what does this agent want and what should I do about it, instead of first running a slow audit of whether the behavior was accidental. Psychologists James Moore and Alicia Pope explain this using error management theory: the two possible mistakes here aren't equally costly. Wrongly assuming an accidental bump was deliberate costs a moment of unnecessary irritation. Wrongly assuming a deliberate shove was accidental could cost you the chance to protect yourself from a real threat. When one kind of mistake is cheap and the other is dangerous, a mind that's tuned to make the cheap mistake more often comes out ahead on average, even though it's "wrong" more often in the narrow, single-instance sense.
Where it misfires
The same default fires just as readily on genuinely accidental, incompetent, or random actions, a stranger's driving, a glitchy app, a friend's bad timing, producing unnecessary conflict, false accusations, and, per research covered in Section 05, a greater pull toward conspiracy-style explanations for events that were never actually planned by anyone.
The live academic debate, briefly
Researchers in the Kahneman/Tversky tradition tend to frame this as an error: substituting a fast, intuitive mind-reading guess for the slower work of actually gathering evidence about what someone intended. Gerd Gigerenzer's ecological-rationality view would treat the underlying default as a sensible, fast and frugal rule for a small-group ancestral environment where most ambiguous actions by people you knew probably did mean something, one that now misfires more often given how much of daily life involves strangers, machines, and pure chance.
How it activates, step by step
1. You observe someone's action, and the outcome affects you, for better or worse.
2. Your mind defaults to reading the action as chosen and goal-directed, rather than first checking for accident, incompetence, or chance.
3. That "on purpose" reading gets treated as an established fact about the world, not a guess.
4. Everything downstream, blame, trust, anger, suspicion, gets built on top of that assumed intent.
Anything that turns an ordinary mistake, coincidence, or complicated outcome into evidence of someone's deliberate plan is exploiting this bias.
A downturn, a shortage, or a run of bad luck gets reframed as something a target group is doing to "us" on purpose, which satisfies the bias's pull toward an intentional cause far more completely than "this outcome has many small, undirected causes" ever could.
An incidental, unremarkable connection to a bad actor gets treated as proof of a deliberately chosen alliance, letting the bare fact of association stand in for evidence of shared intent that was never actually shown.
Tension between two groups gets presented as one side's deliberate, targeted campaign against the other, activating the bias's threat-detection reflex directly, rather than describing what's often a messier outcome of many smaller, less coordinated causes.
A hostile label, "saboteur," "traitor," "wrecker," assumes deliberate intent as a settled fact before any has actually been shown, and every action the labeled person takes afterward gets filtered through the assumption that they meant it.
Outside politics
When a company has an outage, a cancelled flight, or a pricing glitch, social posts blaming it on the company doing this "on purpose to save money" routinely outperform the more accurate, less satisfying explanation, an ordinary operational failure, because the intentional version supplies someone to be angry at and the accidental version doesn't.
What increases it
Strong: schizotypy, a spectrum of schizophrenia-like traits present to some degree in the general, non-clinical population. Moore & Pope (2014), using video clips of ambiguous hand movements rather than written sentences, found that higher schizotypy scores predicted a stronger intentionality bias specifically for socially ambiguous actions, not for physical or mechanical ones.
Moderate: alcohol intoxication. In a controlled taste-test study, Bègue, Bushman, Giancola, Subra & Rosset (2010) gave men either alcoholic or non-alcoholic cocktails, then had them judge 50 ambiguous sentences. Intoxicated men judged the actions intentional 43% of the time, compared to 36% for sober participants; expectancy alone (believing you'd been given alcohol when you hadn't) did not produce the same effect.
What does NOT cleanly predict it
Rosset's original account predicted that taxing working memory should increase the bias, on the theory that concluding "accident" is the effortful, override step. That specific prediction is now genuinely contested: Eisenkoeck, de Fockert & Moore, across two directly designed experiments, found no effect of working-memory load on intentionality judgments, and concluded Rosset's dual-process model is "incomplete and needs to be revised." This is a live, unresolved disagreement about the underlying mechanism, not a settled finding either way, and the page won't smooth over it.
State-dependent factors
Alcohol is the one state-dependent factor directly tested and confirmed above. Acute anger or a felt sense of threat plausibly works the same way given the error-management framing in Section 02, a mind already primed for danger should lean harder toward the "cheap mistake" of assuming hostile intent, but that specific interaction wasn't directly tested in the sources reviewed for this page, so it's noted here as a reasoned extension of the theory, not an established finding.
From a cut-off in traffic, to a lab sentence about a house fire, to a fully formed conspiracy theory, the same reflex keeps showing up: fill in the blank with "on purpose" before checking whether it's true.
First named
"Intentionality" as a philosophical term, the mind's capacity to be "about" or "directed at" something, goes back to Franz Brentano's 1874 work in philosophy of mind. The specific, empirically testable bias covered on this page is a much later and much narrower idea: psychologist Evelyn Rosset isolated and named it in 2008. A related but distinct thread, Fritz Heider and Marianne Simmel's classic 1944 study showing that people readily attribute purposeful, social behavior to simple moving shapes with no minds at all, established decades earlier that humans are quick to perceive agency and intention in general; Rosset's specific contribution was showing the same default operates on judgments of real people's real, describable actions.
The landmark study
Rosset's 2008 paper, "It's no accident: Our bias for intentional explanations" (Cognition, 108(3), 771-780), ran three experiments. The first showed participants judged more actions to be intentional under time pressure than with unlimited time to think. The second showed people spontaneously described even prototypically accidental scenarios in intentional terms when simply asked to explain what happened. The third showed that judging an action as unintentional took more cognitive effort than judging it intentional, evidenced by better memory for the unintentional items. Together, Rosset argued, these point to intentional interpretation as the mind's automatic first guess, with "accident" requiring extra work to conclude.
Key evolution
Bègue, Bushman, Giancola, Subra & Rosset (2010) extended the finding to alcohol intoxication. Moore & Pope (2014) extended it with a video-based methodology and linked bias strength to schizotypy, proposing an error-management-theory account (Section 02) in place of Rosset's original framing. Brotherton & French (2015) extended it again, to conspiracist belief.
Contested and connected findings
The mechanism Rosset originally proposed, that limiting cognitive resources should push judgments toward "intentional" because concluding "accident" is the effortful step, has not held up well. Eisenkoeck, de Fockert & Moore ran two experiments directly testing this prediction with a working-memory load and found no effect either time, concluding the dual-process account "needs to be revised" and that a single-system explanation might fit the data better. The existence of the bias itself is well replicated across multiple labs and methods; the specific dual-process explanation for why it happens is the part that's now genuinely in question.
Score: 1 out of 5 (see the breakdown in the card after Section 01).
The in-the-moment question
"Have I actually ruled out accident, mistake, or something I don't know about, or did I just skip that step?" A reading built on real evidence of intent survives that question. A default reading usually doesn't, because there was never any evidence gathered to begin with.
Why this is hard to catch
There's no discrete moment where the intentional reading arrives after the accidental one and elbows it aside; Rosset's own speeded-judgment result suggests the intentional reading tends to arrive first and faster, not second. It doesn't feel like a conclusion you reached, it feels like part of what you saw, which leaves no natural pause point where a person would think to double-check it.
Debiasing research note
The research reviewed for this page mostly documents the bias and what strengthens it, schizotypy, alcohol, a contested claim about cognitive load, rather than testing interventions that reliably weaken it. There isn't a well-established, directly tested debiasing technique specific to intentionality judgments in the sources used here; the individual and systemic moves above are reasonable extensions of the mechanism, not claims backed by a dedicated intervention study, and this page won't pretend otherwise.
Commonly co-occurs with
Often mistaken for
Not the same question. Fundamental attribution error explains someone's behavior by their character rather than their situation, even for behavior nobody thinks was chosen on purpose, a stranger tripping on uneven pavement gets read as "clumsy," not as someone who meant to trip. Intentionality bias is the earlier, narrower question: was this action chosen at all. The two often chain together, intentionality bias decides "she meant to do that," then fundamental attribution error decides "and that's just the kind of person she is."
A related research area, studied mainly in work on aggression, without its own Filter page yet. It's a narrower case: specifically reading ambiguous actions as hostile or malicious, not just intentional in general. A person can show a strong general intentionality bias while still being fairly accurate about whether an intent, once assumed, is hostile or benign; hostile attribution bias is specifically about that malice reading.
Teleological thinking doesn't require any agent to be named at all, a mountain or an illness can get treated as having a "reason" without anyone having designed it on purpose. Intentionality bias always needs one specific agent's one specific action; it's the narrower, agent-requiring case.
Compounding effect
Intentionality bias often feeds fundamental attribution error directly, deciding an act was on purpose is frequently the first step toward deciding it reflects someone's character, and it can feed just-world hypothesis too: once an outcome looks deliberate, it's a short step to deciding the person on the receiving end somehow deserved it.
Primary sources
Accessible reading