Agency Detection Task
Also known as: HADD Experiment, Agency Detection Bias Task, Hypersensitive Agency Detection Device Test, Agency Attribution Paradigm
The agency detection task is an experimental method in the cognitive science of religion that measures the human tendency to attribute ambiguous events to intentional agents - a tendency Justin Barrett named the Hyperactive (or Hypersensitive) Agency Detection Device, or HADD. Building on Stewart Guthrie's argument that people anthropomorphize the world, Barrett proposed in 2000 that an evolved bias to err on the side of detecting agents (better to mistake the wind for a predator than the reverse) provides a natural cognitive foundation for belief in gods, spirits, and ghosts. The task presents participants with ambiguous motion, sounds, or images and uses signal-detection theory to separate genuine sensitivity to agents from a liberal response criterion, then relates the resulting over-detection bias to supernatural belief.
Key highlights
- Operationalizes a core cognitive-science-of-religion hypothesis - hyperactive agency detection - in a precise, testable behavioral measure.
- Uses signal-detection theory to separate genuine sensitivity from response bias, so an over-detection claim is not confounded with perceptual ability.
- Allows experimental manipulation of threat, uncertainty, and priming to test when the agency bias intensifies, sharpening the causal account.
- Links a laboratory measure to a substantive theory of the natural foundations of belief in gods, spirits, and ghosts.
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Use the agency detection task when you want to test cognitive-science-of-religion claims about why belief in intentional, unseen agents comes naturally to humans, or to measure individual and situational differences in the propensity to over-attribute agency. It is appropriate when you can construct stimuli whose true agency status is controlled and near-ambiguous, collect responses you can score against ground truth, and apply signal-detection analysis to separate sensitivity from bias. The task is well suited to studying how threat, uncertainty, control deprivation, or priming shift the criterion, and to relating that shift to supernatural belief, anthropomorphism, or paranormal ideation. It is less appropriate when stimuli are unambiguous (so there is no room for bias to operate), when ground truth cannot be defined for each trial (precluding d' and c), or when the research question concerns the doctrinal content or social organization of religion rather than its perceptual-cognitive triggers.
Strengths & limitations
- Operationalizes a core cognitive-science-of-religion hypothesis - hyperactive agency detection - in a precise, testable behavioral measure.
- Uses signal-detection theory to separate genuine sensitivity from response bias, so an over-detection claim is not confounded with perceptual ability.
- Allows experimental manipulation of threat, uncertainty, and priming to test when the agency bias intensifies, sharpening the causal account.
- Links a laboratory measure to a substantive theory of the natural foundations of belief in gods, spirits, and ghosts.
- Establishing the true agency status of each ambiguous stimulus can be artificial, and a poorly calibrated ground truth distorts both d' and c.
- A liberal criterion in the lab need not reflect the agency attributions people make about real misfortune, fortune, or natural events.
- Evidence linking measured agency bias to supernatural belief is mixed, and effect sizes are often small and moderated by culture and context.
- The HADD account explains why agent concepts are easy to entertain but not, by itself, why specific gods are believed, worshipped, or sustained socially.
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
What is the difference between agency detection sensitivity and bias?
Sensitivity (d') measures how well a person actually tells apart situations that contain an agent from those that do not - their perceptual acuity. Bias, captured by the criterion c, measures how willing they are to say 'agent' regardless of acuity. The HADD hypothesis is specifically a claim about bias: people are predicted to adopt a liberal criterion that over-produces agent reports, not necessarily to be more accurate. This is why signal-detection analysis is essential - it lets researchers show an over-detection tendency that is independent of how good someone is at detecting real agents.
Does a hyperactive agency detector prove that religion is an evolutionary accident?
Not on its own. HADD is offered as a proximate cognitive mechanism that makes belief in unseen agents intuitive and easy to acquire; whether religion is a byproduct of such mechanisms or has been shaped by selection for social or other functions is a separate, debated question. The agency detection task measures the bias and its correlates; it does not by itself adjudicate the broader evolutionary debate. Barrett and Boyer present HADD as one of several inference systems contributing to the naturalness of religious concepts, not as a complete theory of religion.
Why must the stimuli be ambiguous?
If a stimulus clearly contains an agent or clearly does not, responses are driven by easy perceptual facts and reveal little about interpretive bias. The agency detector is theorized to matter precisely under uncertainty - the rustle that could be wind or predator. By setting the true probability of agency near chance, the task forces participants to fall back on their default leanings, exposing whether they tilt toward 'something is there.' Ambiguity is therefore not a nuisance but the experimental condition under which the predicted over-detection bias can be observed and measured.
Sources
- 1.Barrett, J. L. (2000). Exploring the natural foundations of religion. Trends in Cognitive Sciences, 4(1), 29-34.
- 2.Boyer, P. (2001). Religion Explained: The Evolutionary Origins of Religious Thought. New York: Basic Books.ISBN 9780465006953
You have read it. What now?
Cite this page
ScholarGate. (2026, June 23). Agency Detection Task. ScholarGate. https://scholargate.app/religious-studies/agency-detection-task