Why Is the Affect Heuristic Significant?

Click here and press the right key for the next slide.

(This may not work on mobile or ipad. You can try using chrome or firefox, but even that may fail. Sorry.)

also ...

Press the left key to go backwards (or swipe right)

Press n to toggle whether notes are shown (or add '?notes' to the url before the #)

Press m or double tap to slide thumbnails (menu)

Press ? at any time to show the keyboard shortcuts

Why Is the Affect Heuristic Significant?

Spoiler: this lecture is about two puzzles. The message will be that it is hard to address both simultaneously.

puzzle

Why do feelings of disgust sometimes influence moral intuitions?

(And why do we feel disgust in response to moral transgressions?)

puzzle

Why do patterns in moral intuitions reflect legal principles humans are typically unaware of?

moral psychology can distinguish

actual motivation

from

avowed motivation

There are multiple kinds of disgust.

Important that each serves an adapative function

pathogen disgust—humans vs vultures (both scavengers, but humans do not have the kind of immune system that lets them munch on internal organs by inserting their heads into the anus of a dead antelope).

Incidentally, contemporary hunter–gatherers still scavenge. Pontzer (2021) describes a visit to the Hadza, a group that live in what is now legally Tanzania. He was in his tent terrified of these lion sounds. When day broke, some of the men from the camp came by to show him a freshly killed antelope. ‘Where did you get that from so early in the morning?’ ‘We took it from the lions.’

pathogen

sexual

injury

(Tybur, Lieberman, & Griskevicius, 2009)

Pontzer (2021, p. Chapter 1)

illustration: gender and toilets

This is both an illustration I will use later, and also something of interest in its own right.

Which factors influence attitudes toward policies that limit the choice of toilet to an individual’s birth sex?

See transcript: two bits of background. Also Vanaman and Chapman focus on pathogen disgust specifically.

Method: measure each participant’s views on harm and sensitivty to pathogen disgust; to what extent do each of these factors predict the participants’ attitudes toward policies?

(Vanaman & Chapman, 2020)

How distguisting is ...

stepping in dog pooh?

seeing mold on some leftovers in your refrigerator?

finding a hair in your food?

0–6 [not at all–extremely distgusting]

(Tybur et al., 2009)

Vanaman & Chapman (2020) measure disgust using the three domains from Tybur et al. (2009): pathogen disgust, sexual disgust and injury disgust (which are adaptive for problems related to pathogen avoidance, mate choice, and social interaction). (They also distinguish moral disgust, which is not considered here)

Vanaman and Chapman consider a variety of different models. Here is one. The point of this particular model is to explore the extent to which disgust is influencing support for bathroom restrictions via concerns about harm. The conjecture under consideration is that disgust might make you more aware of issues around harm and that could then explain why you support bathroom restrictions.

But the results do not support this conjecture. ‘the indirect path via purity concerns was over three times as large as the indirect effect via harm concerns (b = –0.08, SE = 0.02, p < .001).’

Complication: What I did not explain earlier is that disgust can have both a direct effect on support for bathroom restrictions and also an indirect effect via concerns about purity. (This is explained by Moral Foundations Theory.)

See Wagemans, Brandt, & Zeelenberg (2018, p. 287): ‘disgust sensitivity is primarily associated with moral judgments in the purity domain.’

‘the indirect path via purity concerns was over three times as large as the indirect effect via harm concerns (b = –0.08, SE = 0.02, p < .001).’

[Here they are considering harm as a mediator because they are considering a hypothesis on which disgust is actually a trigger for harm. But they also looked at the direct effect of harm on support for bathroom restrictions.]

Vanaman & Chapman (2020, p. figure 3)

illustration: gender and toilets

Which factors influence attitudes toward policies that limit the choice of toilet to an individual’s birth sex?

Method: measure each participant’s views on harm and sensitivty to pathogen disgust; to what extent do each of these factors predict the participants’ attitudes toward policies?

Results: Your attitude towards policies that limit choice of toilet
is mainly a consequence of disgust (mediated by purity), not concern about harm.

‘despite the prevalence of harm rhetoric among those who support bathroom restrictions (Fernandez & Blinder, 2015), opinions are probably more strongly driven by moral concerns rooted in disgust than by concerns about possible harm toward either cisgender or transgender people.’ (Vanaman & Chapman, 2020, p. 210)

policy implication:

Intervention: target disgust, not concerns about harm.

‘it would be relatively less helpful to point out that open bathroom access does not result in harm toward cisgender women and girls [...] Instead, interventions (and arguments) that directly target disgust and disgust-driven moral concerns are likely to have a larger impact.’ (Vanaman & Chapman, 2020, p. 211)

So it might have seemed like there are two sets of clashing ethical concerns and we should find balance between them ...

But if Vanaman & Chapman (2020)’s results are correct, the problem is that, as a society, we are just not queer enough.

(Vanaman & Chapman, 2020)

moral psychology can distinguish

actual motivation

(e.g. feelings of disgust)

from

avowed motivation

(e.g. potential harms)

the gap

This is an illustration of the gap.

New tech: bathrooms (very recent innovation) (You might not remember this but it was well into the 1800s before many of us were using toilets—and even today there are plenty of places where indoor toilets, let alone bedroom toilets, are rare.)

New tech: farming allowed males to dominate female reproductive strategies to an entirely novel extent, which has had a big impact on cultural notions of gender.

Old Ethics: driven by pathogen disgust.

‘the central phenomena are moral emotions and intuitions.’

(Haidt, 2008, p. 65)

What does central mean? It’s always a metaphor. Here I think Haidt has in mind that these are the foundations on all other ethical abilities are built.

Moral emotions and moral intutions can exist independently of any other moral abilities; but no other moral abilities could exist if these did not.

We asked two questions about moral intuitions

Q2 What do humans compute that enables them to track moral attributes?

Q1 How, if at all, do emotions influence moral intutions?

What is a heuristic?

In this context we are using heuristic in a special technical sense.

A heuristic links an inaccessible attribute to an accessible attribute such that, within a limited but useful range of situations, someone could track the inaccessible attribute by computing the accessible attribute.

What is an attribute? I’m 178cm tall and a bit round. Those are both attributes of me.

Imagine someone kicking a baby seal. That is a bad action. The badness is an attribute of the action.

What is it for an attribute to be accessible or not?

‘We adopt the term accessibility to refer to the ease (or effort) with which particular mental contents come to mind.’

Kahneman & Frederick (2005, p. 271)

Stress the link to speed–accuracy trade-offs (less effort so faster).

NB: whole brain energy consumption does not vary a lot when thinking hard vs resting the brain. But clearly the brain cannot be used for too many difficult things simultaneously (and different cognitive activties involve changes blood flow and glucose utilization in different parts of the brain). ‘Attempts to measure whole brain changes in blood flow and metabolism during intense mental activity have failed to demonstrate any change (2).’ (Raichle & Gusnard, 2002, p. 2)

device	What is tracked? (accessible)	What is computed? (inaccessible)
poison detector	toxicity	how encountering it makes me feel
moral intuition	right and wrong	how disgusted it makes me feel

What we have learned so far is that there is some reason for thinking that disgust is supposed to work like this. Why do you feel pathogen disgust, sexual disgust or injury disgust? At root, it is because these feelings have enabled humans to avoid dangerous situations.

heuristics are a key feature of cognition

Philosophers neglect these because they are not so interested in how minds actually work.

The use of heuristics seems superficial but turns out to be a revealling feature of much cognition (human and other).

cannonical example

frequency and risk judgements

(Pachur, Hertwig, & Steinmann, 2012)

VSL [omitted here for simplicity]: ‘we asked people to indicate for each risk the value of a statistical life (VSL), which refers to the cost of reducing the number of deaths in a specific class of risk by one.’

dread score: how much participants would dread dying from that cancer. ‘The dread score for each risk was calculated as the mean rating on the 12 characteristics assessed in the risk questionnaire, coded such that a higher value indicates higher dread. ... In a risk questionnaire, people were asked to rate the 24 types of cancer on the 12 risk characteristics that Slovic, Fischhoff, and Lichtenstein (1980) found to contribute to the dread factor: voluntariness of risk, preventive control, control of severity, chronic-catastrophic, common-dread, certain to be fatal, equity of risks and benefits, threat to future generations, personal exposure, potential of global catastrophe, changes in risk, and ease of reduction.’

availability: how many people did each participant have in their social networks that were affected by the kind of cancer in question?

Judgements about whether an action a cancer is good or bad

Note: the attributes to be tracked (frequency and risk) are inaccessible; and the correct answers are always identical.

People in the US tend not to know much about Switzerland. Probably best to keep it this way—don’t want Switzerland going the way of Greenland and Canada.

Pachur et al. (2012)’s conjecture

Participants will use different heuristics for frequency and risk.

The hypothesis is that participants will use these heuristics

frequency

Availability Heuristic

The more easily you can bring cases of it to mind, the more frequent it is.

`it` is the cause of death

(Not mentioned: role of metacognitive processes in monitoring fluency)

risk

Affect Heuristic

The more dread you feel when imagining it, the more risky it is.

`it` is the cause of death

Barak Obama line: more risk of dying from slipping in the bath than from terrorism; but the latter causes much more dread

‘Across Studies 1 and 2, we found that affect associated with risks can be conveniently and parsimoniously gauged by a single item asking for the amount of dread a risk evokes. This is a very useful observation for future studies on the role of affect in risk judgments. [...By contrast,] availability-by-recall offered a substantially better descriptive account than the affect heuristic when people judged deindividualized, statistical mortality rates’ (Pachur et al., 2012, p. 324)

results: predictions broadly confirmed

compare methods: risk judgements vs ethical judgements

The common feature is that you want to show that the judgements are determined by an accessible attribute (cases in social network, feelings of disgust)

Because the idea of a heuristic is so important, I’ve just been giving you examples of two heuristics (Availability and Affect).

Cost: accuracy (and therefore risk of bias)

Should not overstate the risk of bias.

Yes participants did wrongly rate the risk of dying from Siphilis higher than dying from a tornado in Switzerland, but the numbers here are extremely small.

But if you look at the table, people’s ranking of frequency of deaths in Switzerland is not that bad given how little information they probably have used in producing it.

Benefit: speed

stess (i) reliability within limits and (ii) bias

In some cases, they're actually doing pretty well at organising causes of death by frequency and risk. It's not terrible at all. There are some funny things. So for example, they think it's about four times more likely that in Switzerland you would die of syphilis than be involved in a tornado and killed that way. But in fact, it's the reverse, right? Tornadoes are the more dangerous thing in Switzerland. Who knew? But those are very, very small numbers. So I think like any of us, even a Swiss person might be forgiven for getting that the wrong way around. So yes, there is bias involved here, and I think you can easily see why that. The syphilis is a much more terrifying thing for most people than the tornado.

So you can kind of easily see why we might get some bias, but on the whole, the heuristics are working really well.

heuristics are a key feature of cognition

This is why I said heuristics are a key feature of cognition. Not just ethics but everywhere you need to gain speed.

‘the central phenomena are moral emotions and intuitions.’

(Haidt, 2008, p. 65)

Q2 What do humans compute that enables them to track moral attributes?

Q1 How, if at all, do emotions influence moral intutions?

Hypothesis 1

The ‘affect heuristic’:
‘if thinking about an act [...] makes you feel bad [...], then judge that it is morally wrong’.

I did just spend a bit of time on heuristics: I assume we know what heuristics are and why they are significant (speed–accuracy trade–offs).

But the Affect Heuristic is significant in at least two further ways.

Implication 1: ‘if moral intuitions result from heuristics, [... philosophers] must stop claiming direct insight into moral properties’ (Sinnott-Armstrong et al., 2010, p. 268).

Illustration for direct insight.

I have my doubts about the argument here. As far as I can see, the argument for the Affect Heuristic depends on moral attributes being inaccessible. So it isn’t a consequence of the Affect Heuristic heuristic that intuitionism is false: instead we would need to show that intuitionism is false in order to know that the Affect Heuristic hypothesis is true.

Implication 2: Should we trust moral intuitions? ‘Just as non-moral heuristics lack reliability in unusual situations, so do moral intuitions’ (Sinnott-Armstrong et al., 2010, p. 268)

(Sinnott-Armstrong et al., 2010, p. 268).

relevance: defending consequentialism

Example is ‘such as that we are morally permitted to punish an innocent person in the well-known example where this is necessary to stop riots and prevent deaths. ’

‘Critics often argue that consequentialism can’t be accurate, because it implies moral judgments that are counter-intuitive, such as that we are morally permitted to punish an innocent person in the well-known example where this is necessary to stop riots and prevent deaths. With the heuristic model in hand, consequentialists can respond that the target attribute is having the best consequences, and any intuitions to the contrary result from substituting a heuristic attribute.’

(Sinnott-Armstrong et al., 2010, p. 269).

illustration: gender and toilets

recall the illustration from earlier

So where was I? Explaining the significance of the Affect Heuristic.

If true, it has real-world consequences for ethics in philosophy and for policy development.

‘the central phenomena are moral emotions and intuitions.’

(Haidt, 2008, p. 65)

Q2 What do humans compute that enables them to track moral attributes?

Q1 How, if at all, do emotions influence moral intutions?

‘ethical philosophers intuit the deontological canons of morality by consulting the emotive centers of their own hypothalamic-limbic system ...
Only by interpreting the activity of the emotive centers as a biological adaptation can the meaning of the canons be deciphered’ (Wilson, 1975, p. 563 quoted in Haidt, 2008, p. 68).

Hypothesis 1

The ‘affect heuristic’:
‘if thinking about an act [...] makes you feel bad [...], then judge that it is morally wrong’.

Implication 2: Should we trust moral intuitions? ‘Just as non-moral heuristics lack reliability in unusual situations, so do moral intuitions’ (Sinnott-Armstrong et al., 2010, p. 268)

So we got a lot out of the Affect Hypothesis.

but is it true?

Why Is the Affect Heuristic Significant?

Lecture 02

Moral Psychology

[email protected]

Keyboard Shortcuts?f

Why Is the Affect Heuristic Significant?

Lecture 02

Moral Psychology

[email protected]

Keyboard Shortcuts`?f`