012 min
Where the affect heuristic shows up
Here is an illustrative case, using a role rather than a person. A designer is in a product review. The team proposes a feature that records which screens each user opens. The designer built the prototype and likes it. Two numbers are on the table: an estimate of the privacy risk and an estimate of the benefit to the product.
The affect heuristic predicts what happens next. The designer reads the feature's name and feels good about it before reading either number. That feeling then shapes both ratings. They judge the privacy risk to be low and the benefit to be high.
Now change one input. The identical feature was proposed by a team the designer distrusts, and the designer feels uneasy about it from the first sentence. The risk estimate and the benefit estimate are the same documents as before. The prediction reverses. The designer judges the risk to be high and the benefit to be low.
The facts did not differ between the two runs. The feeling at the start did. The moment the effect operates is early: it is the instant a person reads the name of the thing, before they open the data.
Researchers call the feeling in this scenario affect. The same pattern is predicted wherever a person must judge something they already have a feeling about: a technology, a vendor, a medical test, a policy.
Liked, so low risk and high benefit. Disliked, so high risk and low benefit.
022 min
Where the affect heuristic comes from
The affect heuristic came from risk research over several decades, not from one experiment. Four steps matter.
1978: dread. Fischhoff, Slovic, Lichtenstein, Read and Combs asked people to rate technological hazards. Slovic and colleagues later summarised the result of those studies this way:
Feelings of dread were the major determiner of public perception and acceptance of risk for a wide range of hazards.
Dread is a strong fear of a hazard, linked to outcomes that feel involuntary, uncontrollable, deadly or unfair. The same studies found that risk and benefit are positively correlated in the world, because risky activities tend to bring larger benefits. In people's minds the two are negatively correlated. A hazard judged as high in risk is judged as low in benefit, and the reverse.
1980: feelings come first. Robert Zajonc argued that affective reactions are often the very first reaction to a stimulus. They occur automatically and then guide later thinking and judgment.
1994: the link. Alhakami and Slovic asked people to rate activities such as using pesticides on scales such as good/bad and nice/awful. The strength of that feeling was linked to the inverse relation between perceived risk and perceived benefit. That result implies that people judge an activity or technology by how they feel about it as well as by what they think about it.
2000: the name. Melissa Finucane, Alhakami, Slovic and Johnson ran experiments that tested the idea directly. A heuristic is a mental shortcut. The label "affect heuristic" comes from Finucane, Alhakami, Slovic and Johnson in 2000, who reasoned that using an overall feeling is a mental shortcut.
Slovic, Finucane, Peters and MacGregor then gave the idea its fullest account in a 2002 chapter in the book Heuristics and Biases, and again in a 2007 article in the European Journal of Operational Research. Their definition of the key word is narrower than everyday speech:
As used here, affect means the specific quality of goodness or badness (a) experienced as a feeling state (with or without consciousness) and (b) demarcating a positive or negative quality of a stimulus.
Affect is therefore not the same as a strong emotion such as rage or panic. It is a quieter signal, and a person may not notice it.
032 min
Why the affect heuristic happens
Images and the affect pool
Slovic and colleagues say that every image in a person's mind has some positive or negative feeling attached to it, with or without awareness. An image here is a mental picture or symbol of something. The authors call the total of these feelings a person's affect pool.
When a person makes a judgment, they consult the affect pool. The authors compare this to two older mental shortcuts. People judge how likely something is by how easily examples come to mind, which is the availability heuristic, and by how much it resembles a typical case. Affect is a third cue of the same kind. It can be used for many judgments, including judgments of probability.
The authors give the reason a feeling gets used:
Using an overall, readily available affective impression can be easier and more efficient than weighing the pros and cons of various reasons or retrieving relevant examples from memory, especially when the required judgment or decision is complex or mental resources are limited.
Two modes of thinking
The affect heuristic belongs to the dual-process theories. These theories describe two ways of understanding reality. One is intuitive, automatic and experiential. The other is analytical, deliberate and verbal. Many writers call the fast mode System 1 thinking. The authors call it the experiential system, and affect is its basis.
They stress that both modes run all the time, and that the fast mode is not a defect. Before there was probability theory, intuition and instinct told people whether an animal was safe to approach or the water was safe to drink. They also argue that analytic thinking is unlikely to work well without guidance from affect, citing the neurologist Antonio Damasio.
What makes a feeling usable
Slovic and colleagues use the word evaluability for how easily information maps onto a good-or-bad scale. A number that is hard to place on that scale, such as "150 lives", evokes little feeling. A number that is easy to place, such as "98 percent", evokes a lot. This idea explains several results below.
When the heuristic gets stronger
Analytic thinking needs effort, attention and memory. Slovic and colleagues write that ill health, stress and time pressure are likely to increase reliance on affect. Only time pressure has been tested directly, in the experiment described next.
042 min
The experiments that tested the affect heuristic
Two experiments in the 2000 paper by Finucane and colleagues are the main evidence for the heuristic. They were designed to separate affect from analysis.
Experiment one: change the information, not the facts. If affect drives risk and benefit judgments, then information about one should change the judgment of the other. Information that says a technology has high benefit should improve the overall feeling about it. A better feeling should then lower the perceived risk, even though the information said nothing about risk. The researchers gave participants four kinds of information for three technologies. The information said either that benefit was high, that benefit was low, that risk was high, or that risk was low. Slovic and colleagues report the outcome:
Because by design there was no apparent logical relation between the information provided and the nonmanipulated variable, these data support the theory that risk and benefit judgments are influenced, at least in part, by the overall affective evaluation (which was influenced by the information provided).
Experiment two: limit the time. The same group asked people to judge risks and benefits under time pressure, which leaves little room for analysis. The inverse link between the two judgments grew stronger:
Further support for the affect heuristic came from a second experiment by Finucane et al. who found that the inverse relation between perceived risks and benefits increased greatly under time pressure, when opportunity for analytic deliberation was reduced.
The two experiments matter for a specific reason. They show that affect influences judgment directly. It is not just a response to an earlier analytic evaluation that people then rationalise.
These are findings about averages across groups of participants. They describe a tendency, not a rule for every person.
053 min
How affect changes the way people read numbers
Affect also changes how people read numbers. Four findings show how.
Imaging the numerator
Denes-Raj and Epstein (1994) offered people a chance to win $1 by drawing a red jelly bean from a bowl. One bowl held a larger number of red beans but a smaller proportion of them, for example 7 in 100. The other held fewer red beans but a better chance, for example 1 in 10. Many people chose the first bowl. Slovic and colleagues report how they explained it:
These individuals reported that, although they knew the probabilities were against them, they felt they had a better chance when there were more red beans.
The authors call the strategy "imaging the numerator". People picture the winning beans and neglect the size of the bowl. The picture of winning beans carries positive affect, and that affect motivates the choice.
The same risk, written two ways
Slovic, Monahan and MacGregor (2000) tested experienced forensic psychologists and psychiatrists. They asked each clinician whether a hospitalised patient should be discharged. The risk of violence was stated in one of two statistically equivalent forms. One form was a frequency: 20 out of every 100 similar patients. The other was a probability: a 20% chance.
Not surprisingly, when clinicians were told that “20 out of every 100 patients similar to Mr. Jones are estimated to commit an act of violence,” 41% refused to discharge the patient.
However, when another group of clinicians was given the risk as “patients similar to Mr. Jones are estimated to have a 20% chance of committing an act of violence,” only 21% refused to discharge the patient.
Follow-up studies suggested a cause. The frequency form produced frightening images of violent patients. The probability form produced a mild image of one person who is unlikely to harm anyone.
Small probabilities that feel large
When the outcome of a gamble is emotionally powerful, its size matters more than its chance. Rottenstreich and Hsee (2001) tested this. Slovic and colleagues summarise their result:
Support for these arguments comes from Rottenstreich and Hsee (2001) who showed that, if the potential outcome of a gamble is emotionally powerful, its attractiveness or unattractiveness is relatively insensitive to changes in probability as great as from .99 to .01.
A second study makes the same point with a feared disease. Kraus, Malmfors and Slovic (1992) compared experts and the public. Expert toxicologists were sensitive to the size of an exposure to a cancer-causing agent. The public, who felt more strongly about cancer, tended to treat any exposure as quite risky.
In the risk and benefit experiments, the feeling moved two judgments together. Here the feeling replaces the arithmetic. A probability should change a judgment in proportion to its size, but when the outcome carries strong affect it changes the judgment very little.
062 min
The affect heuristic and psychic numbing
Affect is weaker for a large group than for one case. Slovic calls the weak response to large numbers psychic numbing. Two studies measured it.
The first is by Fetherstonhaugh, Slovic, Johnson and Friedrich (1997). They asked respondents how many lives a medical research institute would need to save to deserve a $10 million grant. Slovic reports the result in his later article on psychic numbing:
Nearly two-thirds of the respondents raised their minimum benefit requirements to warrant funding when there was a larger at-risk population, with a median value of 9,000 lives needing to be saved when 15,000 were at risk, compared to a median of 100,000 lives needing to be saved out of 290,000 at risk.
Respondents treated saving 9,000 lives in the smaller group as worth more than saving ten times as many lives in the largest group. The count of lives was not what drove the judgment. The proportion of the at-risk group was.
The second study is Slovic's own airport safety test. Students saw either a measure that would save 150 lives or a measure that would save 98% of 150 lives at risk. The percentage version gained more support. Lower percentages of 95%, 90% and 85% also beat the plain count of 150. Slovic gives the reason in affective terms:
Saving 150 lives is diffusely good, and therefore somewhat hard to evaluate, whereas saving 98% of something is clearly very good because it is so close to the upper bound on the percentage scale, and hence is highly weighted in the support judgment.
This is evaluability at work. A percentage near its maximum is easy to feel as good. A count of lives has no obvious top. The lesson is narrow: do not assume that a large number, by itself, will move people in proportion to its size.
072 min
The affect heuristic in practice
The heuristic shows up wherever a team or an audience must judge risk or benefit.
Risk communication and health
Slovic and colleagues apply the idea to smoking. They argue that many beginning smokers do not weigh risks at all:
Instead, they are driven by the affective impulses of the moment, enjoying smoking as something new and exciting, a way to have fun with their friends.
If affect drives the choice, a message that creates negative affect may work better than a list of statistics. The authors point to graphic warning labels that Canada put on cigarette packs. Hammond and colleagues studied Canadian smokers, and stronger emotional reactions to the labels were associated with more attempts to quit or cut down. A pack from the Philippines with graphic warnings is shown below.
Product and design decisions
An illustrative case for a product manager. A security review lists a small chance of data loss. The study of clinicians above suggests that "1 in 200 accounts" will read as more dangerous than "0.5%", because the first form brings an image of affected users to mind. The two forms state the same number.
The lesson is to choose a format deliberately. Slovic and colleagues note that stars, or words such as "excellent" and "good", make numbers more evaluable. People can then map the numbers onto a good-or-bad scale and use them. The authors also raise the ethical question of when this kind of formatting becomes manipulation.
Decisions about technology
The heuristic also helps explain why public attitudes toward a technology can differ from expert estimates. The study of toxicologists and the public above is one case: the experts followed the size of the exposure, and the public followed the feeling about cancer.
081 min
How to guard against the affect heuristic
A person cannot switch affect off. The guards below reduce how much it controls a decision. They are inferences from the research, not tested procedures, so treat them as sensible defaults.
Remove time pressure from decisions that matter.
The experiment by Finucane and colleagues found that the risk-benefit link grew stronger when people had little time. Give a decision with real consequences a deadline that allows analysis.
Judge risk and benefit separately, in writing.
Ask each reviewer to estimate risk first, then benefit, before any discussion. If every reviewer's risk score falls when their benefit score rises, affect may be driving both.
Notice your first reaction, then ask for the numbers.
Write down how you feel about the option before reading the data. If the data later disagrees with your feeling, treat that as a signal to slow down.
Show both formats for any risk.
Present a risk as a frequency and as a probability, and compare how the audience reads each. The clinician study shows the two can produce different decisions.
Give counts a base.
A bare count like "150 lives" is hard to judge. Add the size of the group it comes from.
091 min
What the affect heuristic does not explain
The authors do not claim the affect heuristic is a flaw to be removed. Four limits matter.
Affect is often correct. Slovic and colleagues argue that the fast mode was essential to survival and is still the most common way people handle risk. A feeling of unease on a dark street is useful information. The heuristic misleads when the feeling comes from something unrelated to the actual risk, such as the way a number is written.
The model is too simple, by its authors' own account. Slovic and Peters (2006) wrote that it is "obviously too simple a model." Two negative emotions, fear and anger, have different effects on risk judgments: fear raises risk estimates and anger lowers them. A single good-or-bad dimension cannot capture that difference.
Most evidence is about risk and benefit. The strongest experiments concern risk, benefit and probability judgments. The studies on psychic numbing and percentages are presented as consistent with an affective account. They were not designed as direct tests against every alternative explanation.
Group results are not individual predictions. People differ in how much they react affectively and in how much they rely on experiential thinking. A person who reasons carefully about a topic can still be influenced, but not always, and not by the same amount.
101 min
The affect heuristic vs nearby concepts
Several concepts in the library are close to the affect heuristic. The axis that separates each one is what the judgment is based on.
- [Availability heuristic](/concept/availability-heuristic). The judgment is based on how easily examples come to mind. Affect is based on how the thing feels. The two often work together, because vivid events are both memorable and emotional.
- [Halo effect](/concept/halo-effect). One good trait of a person or product raises ratings of unrelated traits. The affect heuristic is broader. It covers any overall good or bad feeling and applies to risks, benefits and numbers, not only to traits of a person.
- [Framing effect](/concept/framing-effect). Equivalent options are judged differently depending on how they are described. The frequency and probability formats above are a framing manipulation. The affect heuristic is one explanation for why a frame works, because the frames carry different feelings.
- [Loss aversion](/concept/loss-aversion). A loss hurts more than an equal gain pleases. That is a rule about how outcomes are valued. The affect heuristic is a rule about which feeling is used as information at all.
- [Anchoring](/concept/anchoring). An irrelevant number pulls a numerical estimate toward it. Affect does not need a number. It works on any judgment where a feeling is available.
?8 questions
Questions people ask
What is the affect heuristic?
Who discovered the affect heuristic?
What is an example of the affect heuristic?
Why do risk and benefit seem opposite in people's minds?
Is the affect heuristic the same as the availability heuristic?
Is relying on feelings always a mistake?
How can I reduce the effect of the affect heuristic?
What is psychic numbing?
§9 sources
Sources
Slovic, P., Peters, E., Finucane, M. L. and MacGregor, D. G. (2005). Affect, Risk, and Decision Making. Health Psychology 24(4, Suppl.), S35-S40, DOI 10.1037/0278-6133.24.4.S35
Slovic, P., Finucane, M., Peters, E. and MacGregor, D. G. (2002). The Affect Heuristic. In Gilovich, Griffin and Kahneman (Eds.), Heuristics and Biases, Cambridge University Press
Slovic, P., Finucane, M. L., Peters, E. and MacGregor, D. G. (2007). The affect heuristic. European Journal of Operational Research 177(3), 1333-1352
Finucane, M. L., Alhakami, A., Slovic, P. and Johnson, S. M. (2000). The affect heuristic in judgments of risks and benefits. Journal of Behavioral Decision Making 13(1), 1-17 (SICI)1099-0771(200001/03)13:1<1::AID-BDM333>3.0.CO;2-S
Show all 9 sourcesShow fewer sources
Alhakami, A. S. and Slovic, P. (1994). A Psychological Study of the Inverse Relationship Between Perceived Risk and Perceived Benefit. Risk Analysis 14(6), 1085-1096
Rottenstreich, Y. and Hsee, C. K. (2001). Money, Kisses, and Electric Shocks: On the Affective Psychology of Risk. Psychological Science 12(3), 185-190
Slovic, P. (2007). "If I look at the mass I will never act": Psychic numbing and genocide. Judgment and Decision Making 2(2), 79-95. The link is the author's January 2006 draft
Slovic, P. and Peters, E. (2006). Risk Perception and Affect. Current Directions in Psychological Science 15(6), 322-325
Zajonc, R. B. (1980). Feeling and thinking: Preferences need no inferences. American Psychologist 35(2), 151-175
