Mar 22 2010

Hyperactive Agency Detection

Something does not seem quite right. The most powerful man in the world, John F. Kennedy, was taken out by a lone nutjob of no previous consequence? A jet flies into the pentagon and yet the expected debris is not visible. And why can’t I see stars in the NASA Apollo moon landing photos?

Some hidden agent must be at work, conspiring to deceive and carry out some sinister plot.

At least that is how our brains are hardwired to think, and some of us more than others. This tendency has been termed the “hyperactive (or hypersensitive) agency detection device” –  HADD – coined by Justin Barrett. Understanding that HADD is an intrinsic part of human nature is part of the core knowledge base of the skeptic.

As a neurologist and a skeptic I am particularly interested in how brain function relates to human intellectual strengths and weaknesses and how knowledge of such helps us to avoid common mental pitfalls. In other words, knowledge of how the human brain works helps us think better – to be more skeptical and avoid error.

Psychologists and neuroscientists in recent years have demonstrated that our brains are hardwired to distinguish things in our environment that are alive from those that are not alive. But “being alive” (from a psychological point of view) is not about biology, but agency – something that can act in the world, that has its own will and can cause things to happen. Sure, this is a property of living things, but that’s not how our brain sort things out. We can perceive agency in non-living things if they are acting as if they are agents.

This is reflected even in our visual system, which separates out visual information into different streams according to the type of information. One division is between information about actions and information about objects. The object stream is also divided into brain regions that deal with inanimate objects and other regions that deal with living things or animate objects. So on a fundamental level our brains treat agents different than objects – from the moment we see them.

Bruce Hood, author of Supersense, goes over in his book the psychological studies that have documented and described the human tendency to think of objects differently than agents. We imbue agents with an essence – a unique living force, even while infants. Objects are just generic things, totally interchangeable. While agents have their own unique essence. Interestingly, children can come to view a favorite toy (a stuffed animal, for example) with the properties of an agent and will treat it like a living thing. This reinforces the notion that the distinction we make is not between living and non-living so much as agent vs object. This  likely also explains why we can watch a cartoon and react emotionally to the characters as if they were real – they are not living, but we see them as agents.

According to Barrett, HADD works in part by detecting any movement that is non-inertial – something which seems to be moving of its own volition. We then assume it is acting with agency and react accordingly. This likely provided an evolutionary advantage – it is better to assume the rustling in the bushes was not the wind but a hungry tiger. So we are descended from hominids who were more paranoid and had hyperactive agency detection, because they were less likely to be eaten by predators.

We can extrapolate from “non-inertial movement”, or movement that cannot be easily explained as a passive reaction to natural forces, to more and more complex “actions.” HADD detects more than movement, it can detect a pattern in otherwise unrelated events, details that defy easy explanation, or consequences that seem out of proportion to the alleged causes. When HADD is triggered we tend to see a hidden agent working behind the scenes, making events unfold the way they do, and perhaps even deliberately hiding its own tracks.

When HADD is triggered and we think we see the hidden agent, it speaks to us in a very primal way. For some people the perception of hidden agency becomes overwhelming, dominating all other thought processes. We know these people as conspiracy theorists. But there is a little conspiracy theorist inside each of us.

Studies have also demonstrated that HADD is more likely to be triggered when a stimulus is ambiguous – therefore it tends to be our default assumption – an object is an agent until we are sure it’s just an object. Also, in situations where we have less control our HADD becomes more active still.

Barrett and others have speculated that HADD is important to the development of religion – where God is the ultimate invisible agent. So far this hypothesis has not been significantly researched, but it does seem reasonable. Seeing natural or random events as the will of an agent is HADD.

HADD also leads to superstition – thinking that there is a cause and effect between unconnected events. The underlying assumption of superstitions is that things happen for a reason (a vague ill-defined reason, but a sense that there is a hidden agency at work). I think it is liberating to understand that rather (as was most elegantly stated on a bumper sticker) “shit happens.”

Skepticism, in many ways, is a filter on HADD. First we have to recognize that our brains are not perfect perceivers and processors of information. There are specific and myriad ways in which the human brain is biased and flawed. Science and skepticism are methods for correcting or filtering out those biases. Skeptics ask themselves – is it really true. We see many patterns, but only some of those patterns represent underlying reality. We need a process to sort out which ones are real – that is science and skepticism.

23 responses so far

23 thoughts on “Hyperactive Agency Detection”

  1. Eternally Learning says:


    This kind of stuff is really the reason why I was so drawn to the skeptical community in the first place. I’m endlessly fascinated with exploring the roots of the human condition and I completely agree with you that the more we find out how our minds work the more capable we are of anticipating our mistakes and avoiding/correcting them. Thanks for a great article and keeping us abreast of this type of thing!

  2. johnc says:


    There absolutely has to be a given amount of redundancy in the things we see, or think we can see.

    Every area of science, even maths, has its need for speculation which is then clarified by investigation, so it’s vital that our minds are capable of seeing more patterns and possible connections than there could actually be.

    I have no beef with 911 truthers or moon landing conspiracists, precisely because a world in which people didn’t ask those questions would be a far darker and boring place, even if those questions seem to be in poor taste.

    The law of large numbers (and history) tells us that the conspiracy believers occasionally get it right, too.

  3. Suraky says:

    Johnc, was the war in Iraq sufficiently entertaining for you?

    It’s certainly well known today that Iraq had nothing to do with 911 … but was a convenient hidden agent as seen by Bush, A lot of people needlessly died because of this.

    But at least it wasnt boring eh?

  4. John – I agree. That is why I said that science and skepticism are filters for HADD, they do not eliminate it. It is OK to see all the patterns, as long as you then sort out which ones are real and which are illusions.

    Problems arise from assuming that all patterns are real and further assuming a hidden agency at work.

  5. Watcher says:

    Problems arise from assuming that all patterns are real and further assuming a hidden agency at work.

    Just like you referenced in text, see Evangelical Christianity and the idea that, “Whenever He closes a door, He opens a window.” It’s not to knock any particular religion, because the idea exists in many religions, but it’s interesting to know there’s a physiological representation known that is the basis of these thoughts.

  6. Draal says:

    I’m a little confused on the definition of HADD. Does HADD encompass both processing of visual information as well as abstract information? Here’s what I’m referring to: Suppose an observer sees a car suddenly hop a curb and come barreling down at him.
    Suppose the observer is blindfolded and a second observer verbally describes the same car approaching.
    Is the processing of information into a sense of danger from something we see versus something that is described to us the same?

  7. daedalus2u says:

    All types of pattern recognition systems suffer from two types of errors, there is the type 1 error, the false positive, and there is the type 2 error, the false negative.

    HADD is a type 1 error, of seeing something that isn’t there. There is a trade-off between these two types of errors, you can have zero type 1 errors if you reject every instance, but then your type 2 error rate is maximized. You can have a zero type 2 error rate if you accept every instance but then the false positive rate is maximized.

    There is a deal of pattern recognition in all forms of communication. The bit count of the data stream is insufficient to convey the information content of language without enormous data compression and decompression at the other end. That decompression is pattern recognition. When the compression and decompression are done seamlessly and with little loss in data, there is good communication with no misunderstanding. When the data compression and decompression do not happen seamlessly, then there is data loss, and the message can be garbled or misunderstood.

    Usually the data compression and decompression happens seamlessly and is transparent to the users, they are not aware that it is happening. They become aware when it does not happen seamlessly, when the error rate goes up, there is miscommunication, and then there are default responses.

    I think one of the default responses to a high error rate in communication, and particularly in non-verbal communication is the triggering of xenophobia. It may not always be that simple, but that may be an important component.

  8. johnc says:


    Not sure what you’re getting at. At all. I didn’t find the iraq war entertaining though.

  9. CKava says:

    Nice article Steve and coincidently Justin Barrett is currently one of my supervisors at Oxford! I’d also recommend that he’d be a great guest for the show sometime as he’s a really good presenter and a very interesting guy.

    I’m personally quite convinced by the evidence for a HADD but if you want to see a semi-alternative critical perspective there is a nice article called ‘What’s HIDD’n in the HADD?’ from the Journal of Cognition and Culture in 2007. The author ends up arguing that it’s intentionality rather than agency that we tend to over detect and that there is greater evidence of an actual will defined neural structure dealing for intentionality. Not sure if I agree but you might want to check it out if you have time.

    Anyway, nice to see my interests and studies dovetailing a bit. I’ll mention the article to Justin in case he misses it.

  10. CKava – thanks. I would like to interview him for the SGU also. Can you ask him for his e-mail address or give him mine (


  11. CivilUnrest says:

    Dr. Novella,

    Have you seen anything to suggest that agency detection is a developed trait in humans? More specifically, is this something that some people could (physiologically) have “more” of than others?

  12. daedalus2u says:

    There has been a fair amount of work in the context of autism in what is called a “theory of mind”. People with autism do have a reduced tendency and ability to attribute “agency” to inanimate objects. The classic study shows that people with autism have a reduced tendency to attribute emotional motivations to animated triangles.

    They exhibit what the researchers called a “deficit” in mentalizing because they could not attribute the “correct” mental states to the animated triangles.

    Is this a type 1 error, or a type 2 error? A false positive or a false negative? What is the “correct” mental state of the animated triangles? Do triangles have a mental state? Are the autistic individuals missing something hat is there, a false negative, or are the neurologically typically developing individuals seeing something that is not there, a false positive?

  13. Rillion says:


    I would say that that in the case of the triangles with behavior that NTs interpret as intentional, it would probably count as a false positive for them. Not a lot of research has been done on how people with autism think about supernatural concepts, but I’ve done a couple of preliminary surveys of people with Asperger’s and there does appear to be a marked difference– it stands to reason that if their theory of mind is different, that would include how they “theorize” about supernatural minds. I talked with Justin Barrett about researching this and he was very enthusiastic about it.

    Justin is indeed an interesting guy. I recommend his book Why Would Anyone Believe in God? to people who are interested in what the field of cognitive science of religion is all about, and want to get an easily accessible and concise survey of the prominent theories in addition to Barrett’s own ideas. Steven’s link on HADD refers to two other scholars, the psychologist Jesse Bering and anthropologist Stewart Guthrie. So far as I can tell, the term “pareidolia,” which is often mentioned on SGU, was coined in a 1994 issue of Skeptical American. So Guthrie didn’t use it in his 1993 book Faces in the Clouds, which was basically an extended treatise on how religion is essentially systematized anthropomorphism. He was the one who pointed out that being hyper-vigilant about agency is evolutionarily adaptive since it’s far better in terms of survival to mistake a tree trunk for a bear than the other way around.

    Jesse Bering’s work has primarily focused on beliefs about the afterlife, but existentialism is a central concern for him and he has even gone so far as to posit an “existential theory of mind” which cannot help but interpret highly unlikely and fortuitous events as caused by some kind of agency which has a purpose for us. He would be a great guest on the show as well, btw.

    Justin is one of the few “big name” people working in the cognitive science of religion who is an unabashed theist– his goal in Why Would Anyone Believe… is to show that religious belief is entirely natural (as loaded as that word is). He believes that if God wants us to know him, then it logically follows that he would give us the kind of brains which are eager to detect him.

  14. Rillion says:

    Skeptical American? Umm, I meant Skeptical Inquirer. Sorry for accidentally conflating magazines!

  15. jaranath says:


    I though it was fairly obvious. He was arguing that the justification for the invasion of Iraq was based on false agency detection. I don’t entirely agree; there were other selling points used on the American public, though I think they were mostly bet-hedging and that 911/WMD was the main one.

    But unless I’m mistaken, his real point is that it’s probably wrong to “have no beef with 911 truthers or moon landing conspiracists,” because these sorts of things have real consequences.

    Now, I think you might argue that they’re victims of their own agency detection and so we shouldn’t criticize…if so, I strongly disagree. From a statistical perspective, any given potential conspiracy theory or bit of pareidolia is likely to attract a decent number of followers, but I would argue that a very large proportion of those potential followers have strong enough critical thinking and skepticism, innate or learned, to check themsleves before embracing the belief if they have a little help from others pointing out the problems. I think public criticsm is necessary for that effect, and I think it helps encourage more learned skepticism in general.

    That aside, I also think such people are deserving of direct criticism for the simple sake of their error. Often they should “know better,” and are failing to apply critical thought they normally do elsewhere, out of various personal biases they are capable of recognizing and correcting for.

    Of COURSE I know they will often, even usually, fail to learn from the criticism. I do it all the time, catching myself only sometimes. Nor do I expect people to burn bridges with friends, family and acquaintances who won’t budge. We have to make judgement calls. But wherever possible, I think we should find a way to challenge such beliefs, both for the potential direct benefit of the believer as well as the indirect benefit of broader society. Put the truth where it needs to be.

  16. daedalus2u says:

    Rillion, I agree with you, that the perception of agency in the triangles is an error of the second type, a false positive. However, if you read the autism literature, the researchers attribute the “error” to the people with autism not detecting the agency of the triangles, and this is called a “deficit in mentalizing”. I think this is the HAAD of the autism researchers seeing anything different than their own thinking style as pathology.

    I think this “deficit” is actually one of the major features (in that it provides positive benefits) of the autism spectrum, which I see as a trade-off of a “theory of mind” for a “theory of reality”.

    Communication with language requires a “theory of mind”, which call the cognitive hardware that converts a mental concept into a data stream of language and back. Fundamentally the only things that can be communicated are mental concepts, the first person has a mental concept, converts it to the data stream of language, the data stream is transferred, the second person up-converts the data to mental concepts instantiated in the second persons brain.

    To understand how someone is thinking is to be able to instantiate the same mental concepts and the same thinking processes that lead to those mental concepts.

    Acquisition of a first language by humans is an interesting process. Children either learn the well-formed language in their environment, or they synthesize a new language, a Creole, from the bits and pieces of the pidgin languages being spoken around them. The compulsion that forces the development of a single mapping of sounds and gestures onto mental concepts as the Creole is formed must be quite profound.

    A “theory of mind” can be completely arbitrary. The ToM that is “correct”, is the one shared by everyone else. A “theory of reality” is completely different. It has to correspond to actual reality, so it must be changeable when it is found to be wrong. A ToM can be completely fixed over a lifetime, a ToR can’t be.

  17. BillyJoe7 says:

    “Rillion, I agree with you, that the perception of agency in the triangles is an error of the second type, a false positive.”

    Or a false negative?

    How are cartoon characters not agents?
    Is there not intention there?
    Not in the actual characters of course, but in the minds of their creators.
    Cannot a particular nasty cartoon character give a child nightmares?

    What about characters in a book?
    Is there not agency there?
    Not in the characters but in their author working through his characters?
    Cannot these characters affect us in a very deeply meaningful way?

    If not, what about the characters in a movie of the book?
    Are they not agents?
    I guess if you think cartoon characters are not agents, then actors likewise would not not agents either.

    (I hope the above does not come off as being too agressive because actually I’m only asking questions here.)

  18. cloudskimmer says:

    If I read this correctly, normal people falsely attribute agency to animated inanimate objects. The correct assignment would be to realize that the triangles are not animate, as autistic people do. But this is not the typical interpretation, so even though the autistic are correct, they are considered abnormal because the non-autistic make this mistake. So normal kids treat their stuffed animals as if they were alive, but autistic kids don’t. At some point, we know intellectually that such things aren’t alive, but can happily suspend disbelief and enjoy the puppet show. It seems to me that it isn’t a question of who is right or wrong, but whether a kid is autistic or not, and since this is a clear way of distinguishing behavior, it can be used to help diagnose autism.

    Since the Iraq war and WMD’s don’t involve this agency/non-agency interpretation, I don’t see what it has to do with the study. The question was what the Iraq government was doing, and there was no question that they were human beings making decisions and carrying out actions; the only question was what were those actions? UFO’s seem to be a classic example of misattribution of agency, when the alien buffs decide that the lights in the sky must be alien-operated spacecraft. Can this research be extrapolated to all human errors, such as the mistakes leading up the war in Iraq? They seem very different.

    The study shows why, when we make a mistake, it is common to say, “I’ve been HADD!” (Sorry, but it’s late, and I’m feeling punchy.)

  19. jaranath says:

    I dunno, cloudskimmer. I’m not sure the public (or government, for those who honestly believed it) acceptance of the 911/WMD argument is specifically HADD or not. But I could see it being so, or related. We often read intent in others’ actions that simply isn’t there, often with influences from our own biases and desires. Regardless of evidence, supporting Al-Qaeda lined up with many people’s expectation of what Iraq would do, as did developing new WMD. And many of us wanted to invade anyway.

    In that context, are the public and government responses HADD or just a more mundane group of logical fallacies? Of course, maybe this isn’t the best case study given it’s heavy political tension…

  20. clgood says:

    Fascinating post.

    It reminded me, if you’ll allow a slight tangent, of the business I’m in. We absolutely depend on Agency Detection in the audience. The definition of animation we use is “to bring to life”, and the main criterion used for “life” is this: Does the audience believe that the character’s actions are the result of a thought process.”

    Of course, being *really* convincing takes a high degree of skill, which our animators have.

  21. BillyJoe7 says:

    I’m not sure if I’ve hit a nerve 4:58 pm or if my suggestion is so ridiculous as to not be worthy of comment. 🙁

  22. cloudskimmer says:

    BillyJoe7: Your suggestion is certainly worthy of comment; I wondered about that myself. Where does one draw the line between the agent and their creation? In the case of a book/cartoon/movie the agent is the creator of those characters. Characters in a book or cartoon, however real to us, are never agents themselves; how can they be when they have no independent abilities/thoughts/feelings/actions. They are simply the puppets of their creators. Movie characters are a tougher call. The original scriptwriter writes the words, but the actor adds their own interpretation and inflection, and the director chooses what to emphasize, changing the meaning of the original, which is clear when we are disappointed when our favorite book makes a crummy movie. So a movie is a collaboration of many individuals, all of whom are agents.
    And are we willing to consider non-human animals as agents? Would this be used as a way of distinguishing animals from humans, and would it be a valid measure?
    What seemed most interesting to me was that when autistic kids don’t attribute agency to dancing triangles or their teddy bears, they are correct. But since non-autistic kids do, it’s one way to distinguish the autistic. So isn’t this about diagnosing autism and what constitutes normal behavior rather than being correct?

  23. BillyJoe7 says:


    “What seemed most interesting to me was that when autistic kids don’t attribute agency to dancing triangles or their teddy bears, they are correct. ”

    I guess so.
    I suppose the solution to this problem is that, when they don’t attribute the agency behind the triangles, they are not correct.
    And this would therefore have to be classified as a deficiency, not an attribute.

    Of course, those who attribute agency directly to the triangles are living in cloud cuckoo land! 😀


Leave a Reply