Conservation of Threat

Here’s some interesting research about how we perceive threats. Basically, as the environment becomes safer we basically manufacture new threats. From an essay about the research:

To study how concepts change when they become less common, we brought volunteers into our laboratory and gave them a simple task ­—to look at a series of computer-generated faces and decide which ones seem “threatening.” The faces had been carefully designed by researchers to range from very intimidating to very harmless.

As we showed people fewer and fewer threatening faces over time, we found that they expanded their definition of “threatening” to include a wider range of faces. In other words, when they ran out of threatening faces to find, they started calling faces threatening that they used to call harmless. Rather than being a consistent category, what people considered “threats” depended on how many threats they had seen lately.

This has a lot of implications in security systems where humans have to make judgments about threat and risk: TSA agents, police noticing “suspicious” activities, “see something say something” campaigns, and so on.

The academic paper.

Posted on June 29, 2018 at 9:44 AM22 Comments


Bob Dylan's Itchy Feet June 29, 2018 10:02 AM

“Basically, as the environment becomes safer we basically manufacture new threats.”

This is no way surprising, because we make stuff up all the time. The most vivid example of this what happens when you stick a person in a sensory deprivation chamber: they will begin hallucination.

Our minds evolved (at least in part) in response to external stimuli and when there is no stimuli the mind invents it as a way of keeping sane.

bcs June 29, 2018 10:20 AM

Sounds like some sort of calibration loop.

It also leads to a possible psychology hack: fabricate things that are seen as threats but where both the thing it self and (importantly) the reaction to it are both cheap and benign.

Chris June 29, 2018 10:26 AM

The language they use to describe their methods in the essay seems like the researchers may have put a thumb on the scale. Participants were asked, “to look at a series of computer-generated faces and decide which ones seem “threatening.” “. Asking “which of these faces seem threatening?” Is a very different question than “do any of these faces seem threatening?” The former implies that the subjects are supposed to find some of the faces threatening in a way that the later does not (indeed, this sort of phrasing is considered problematic when used in police lineups).

Chelloveck June 29, 2018 11:21 AM

@Chris: It’s possible the question was leading; I’d like to read the full paper (which is behind a paywall) to see their methodology. The article talks about another study they did asking participants to decide whether dots were blue or purple. That one had similar results and on the surface seems to be less likely to play into emotional bias, but the wording could have been leading there, too. “Identify the blue dots” implies that there will be blue dots to identify; “Tell me whether these dots are purple or blue” is much more neutral. The essay does talk about how they tried to control for the participants’ expectations in the color experiment. Hopefully they employed similar controls with the faces.

Martin June 29, 2018 11:27 AM

I don’t have any scientifically backed data, but I’m convinced that organizations established to manage / respond to threats are motivated to have continued or, better yet, publicize the potential of increased threats so their existence will continue or expanded. I know of no organization that has ever said their services were not longer needed because the source of their creation has been eliminated by their actions.

vas pup June 29, 2018 11:40 AM

@Bruce:”This has a lot of implications in security systems where humans have to make judgments about threat and risk”.

I guess that is good article on how to make judgments about threat and risk as well and resist pressure:
“Not surprisingly, social vigilantes score high in dogmatism – the tendency to be closed-minded. But not all closed-minded people take it upon themselves to impose their views on others. Social vigilantes are not only dogmatic but are also highly motivated to control other people, and they narcissistically believe that their views are so incontrovertibly superior that they should make an ongoing effort to change others’ “ignorant” beliefs. Ironically, they are also the sort of people who display a great deal of resistance (what psychologists call reactance) when other people try to persuade or control them which, of course, is what social vigilantes try to do to the rest of us.

Social communities must have standards regarding appropriate behavior to protect people from being disadvantaged and harmed. But, in a complex, heterogeneous society — particularly one that values personal autonomy and freedom of speech — trying to get everyone to think and act in line with one’s own beliefs is not only futile but also arrogant and disrespectful. Social vigilantes display an egoic lack of perspective-taking in thinking that everyone should share their beliefs and that their views should take precedence over everyone else’s.

When I encounter a social vigilante who is determined to impose his or her beliefs on others, I’m reminded of how we responded to bossy kids in elementary school who insisted that everyone do things their way: “Who died and made you sheriff?”

Tatütata June 29, 2018 11:44 AM

Earlier research in threat classification was provided by Jones, Smith, et al in Men in Black, Columbia University Press, 1997

As to myself, I tend to see the world in increasingly threatening shades of orange…

Bernie June 29, 2018 11:46 AM

I’m with the others that want to know the details. “Which ones seem threatening?” implies that some should be threatening. “Are any of these threatening?” would be better, but not good enough. Balance the one end with the other via opposites (eg, safe vs threatening) and vary which word you start with when asking. Or find a single word/term that handles the whole spectrum. There may be even better ways of asking or measuring.

Alyer Babtu June 29, 2018 1:52 PM

This seems parallel to the way sensory systems tune out sameness, like fixed visual and sound patterns, perhaps so that the finite energy budget can be available to respond to novelties. It’s what you want. But above the senses at the level of intellectual knowing, these data are combined with experience and broader background understanding to extract the truth in the situation. Not that we always get there. Stephen Grossberg’s work on neural modeling seems worth looking at in regard to these issues.

echo June 29, 2018 3:52 PM

Other studies have demonstrated that politicians are extremely sensitive to any form of perceived risk and also that board level executives in large companies have minds which essentially have a politicial psychology (i.e. have turned to porridge). Taken together with this study on imaging risk in safer environments what would this indicate?

VinnyG June 29, 2018 4:23 PM

I’d also like to see the unredacted paper. However, even if the “threatening faces” question was leading, the point remains that when faces that were designed to be threatening were removed from the test set, the volunteers apparently continued to find a consistent number of faces to be threatening. Possibly there is something hardwired in the “old” brain that has a certain capacity for a catalog of threats, and tends to keep that catalog “full,” ordering entries from most to least (but never “un”) threatening. Interesting post, as I’m curently reading Annie Duke’s “Thinking In Bets.” which has quite a lot of discussion about overcoming the confirmation bias and related reactions that seem to be burned into the un/subconscious mind.

Clive Robinson June 29, 2018 4:30 PM

@ Bruce,

Basically, as the environment becomes safer we basically

You’ve hit one of my pet peeves that make me wince when I am proof reading my writting like “thus…thus”, “likewise…likewise”, “because…because” and many others such as those starting consecutive paragraphs like “Further… Further”… You have my sympathy because it takes a long time to get rid of those darn doubles, that if you don’t makes things read like faux legalese.

echo June 29, 2018 4:37 PM

So what you’re saying, essentially, is, actually, because…? I had an attack of too many commas once. It’s took ages to shake off.

PeaceHead June 29, 2018 4:39 PM

@Chris, Martin, Babtu: good points, especially about the bias of the experimenters.
@echo, good question.

@Bruce: Really excellently interesting topic. Thanks.

I saw a documentary somewhere that once quoted some former Cold War general as saying something like, “Since there hasn’t been any recent trouble or hostility for such a long time there must be something hiddeously horribly terribly troublesome and hostile happening in the other country and it must be aimed at us in this country. ”

It’s classic paranoia, a psychology topic I’ve studied a lot.
There unfortunately seem to be a lot of people who still think like that.
I worry that White House guy John Bolton is one of those people, who, to quote the spacetime traveler from the movie Dune “sees fires within fires and wheels within wheels”.

The word for that is Apophenia, cognitively considering to have perceived something significant manifested as something or within something that others say is meaningless or not significant at all. For example, seeing coded messages in the patterns on the turf left by the regular ebb and flow of the tides.

The human brain has a tendency to “fill in the gaps” of missing info is what I was taught in a psychoacoustics course in college. I think this topic, as others alluded to, has some parallels. There’s that colloquialism, “people see what they want to see”.

very nice.
Peace is yours.

Clive Robinson June 29, 2018 5:11 PM

@ Chris,

The language they use to describe their methods in the essay seems like the researchers may have put a thumb on the scale.

Yup, there are people that make exceptionaly good livings doing this sort of “Miller’s Thump” trick. You will see them working in “pollster” type organisations and in more modern parlance they are generating “Fake news”…

In the UK certain “Red Top” newspapers do this sort of thing combined with “premium rate phone lines”. One in particular puts a fair degree of “profit” both “Political and Financial” into the hands of a large US News Corp…

Others are even craftier, they have “test questions” before the “poll questions” and give some kind of nonsense about it being a “check / verifier”. The reality is what you say to what appear unrelated questions has a couple of purposes, traditionaly they would be used to select which set of “random poll questions” you got, that were tailored to change your viewpoint progressively or push you into certain answers. But in more modern times they are used to build a marketable databse that pulls in a nice little income.

Have a look at some of the activities of Cambridge Anslytica and Brexit to see the scope of just how bad have got in the “Fake News” front. They had thumbs that weighed in on the scales more than Mnt. Everest ever could…

The amount of money floating around in this Faux Poll/News environment is large, very large. Because if you get it right the person paying you ends up running a country, where “disappearing” people and assets will repay many times over any expediture they might make in your direction. Or in the supposadly slightly more democratized countries (WASP etc monied plutocracies) you might be ensuring that your chosen candidate who is going to give multi-trillion tax breaks and other lucrative legislation gets their “A55 on the throne”…

Alyer Babtu June 30, 2018 12:33 AM


brain has a tendency to “fill in the gaps

and @several others

The sensory system is constantly (nonlinearly) interpolating from incomplete noisy data to maximize signal to the higher levels of the nervous system. It’s a necessity for the excellence of the system, complementary to the filtering out of static data. At the final level, choice and will adjudicate, maybe badly.

Grossberg’s recurrent dynamical system based neural models can by tweaking weight parameters simulate emergent healthy and neurotic behavior.

Otter June 30, 2018 6:11 AM

Did they run a similar experiment asking which faces seem “friendly”?

If they believe their own stated outcome, how can they claim to have designed a range of faces from very intimidating to very harmless?

Frank Wilhoit June 30, 2018 4:06 PM

Other commenters have riffed off of the tag line “…Basically, as the environment becomes safer we basically manufacture new threats….”, mostly analogizing this phenomenon to the manner in which the human sensory apparatus simulates bandwidth that it doesn’t have.

This is a false analogy. “Threat conservation” takes place at a much higher cognitive level and represents a crucial failure of education. It has this in common with most of the other manifestations of severe emotional regression that make up modern politics.

albert July 1, 2018 10:09 AM

Academic studies in psychology are a waste of time. Can’t we find meaningful employment for these people? See:

Kudos to those of you who see through the BS, with some cogent and perceptive observations.

The perception of threats by viewing “computer generated” faces has no analog in the real world. Given the power of social propaganda, and the milieu created by the MSM and it’s corporate controllers, it’s virtually impossible to apply lab studies to the population in general.

“…This has a lot of implications in security systems where humans have to make judgments about threat and risk: TSA agents, police noticing “suspicious” activities, “see something say something” campaigns, and so on….” Like the lady who saw a brown person on the plane writing in a strange language?

This has -no- implications for security systems, except to the degree that the idiots running them have (or lack) common sense.

“…@Bruce: Really excellently interesting topic. Thanks…”

Guess I’ll keep that salt shaker handy.

. .. . .. — ….

Peter July 7, 2018 11:23 AM

Humans have a strong tendency to see what they want to see, or to see what they are told they are seeing. Furthermore, when you constantly run around with a hammer in your hand, everything appears to be nails – hence, you have TSA-fascists performing body-cavity searches on toddlers …. And despite falling crime-rates, you have people screaming for “more security” . This alone makes “democracy” a rather risky affair, IMO we are way beyond the “mob-rule” point .

Leave a comment


Allowed HTML <a href="URL"> • <em> <cite> <i> • <strong> <b> • <sub> <sup> • <ul> <ol> <li> • <blockquote> <pre> Markdown Extra syntax via

Sidebar photo of Bruce Schneier by Joe MacInnis.