Category: Reasoning

ABC Radio National’s All in the Mind just broadcast an interesting interview with behavioural economist Dan Ariely, where he discusses some of his fascinating work on our cognitive biases and why we find it so difficult to judge what will benefit us most.

I’m pretty sure it’s a repeat, but I mention it as I’ve almost finished the unabridged audiobook of his recent bestseller Predictably Irrational which is thoroughly excellent.

The first thing that strikes me is ‘wow, you’ve done so much interesting research’, as the book is largely about studies he has personally been involved with.

The second thing is ‘damn, I wish I’d thought of that’ as the studies are often cleverly conceived and tackle real-world corners of our reasoning and judgement.

The chapters on anchoring and on decoy options are particularly fascinating and he gives a vivid example of how decoy options work.

He notes that the UK magazine The Economist was offering a web only subscription for $59, a print subscription for $125 dollars, and a print-and-internet subscription also for $125.

It seems no-one would choose the print-only subscription – it seems obsolete – but its mere presence affects our reasoning and boosts the sales the more expensive option.

In a study to test this, Ariely gave participants the choice between these three subscription options, and to another group of participants, the choice only between web-only and print-and-internet subscriptions.

in the three option condition 16 people chose the internet-only subscription, none the print-only subscription and the other 84 chose the print-and-internet option.

As the print-only is obselete, it should make no difference whether it is part of the choice or not, when it isn’t there, in the two choice condition, the reverse pattern emerged. The majority, 68 people, chose the cheaper online option, while only 32 took the print-and-internet option.

In other words, the print-only is a decoy and it makes us think that the print-and-internet option is a better deal because it has something ‘free’, when in reality, this impression is just created because we’ve just been presented with a decoy worse deal

This relates to one of Ariely’s main points that he returns to throughout the book, that we tend to make relative judgements, and manipulating the context can skew our perceptions of value.

It struck me that this is how most people experience pitch and musical notes. A few people have ‘perfect pitch‘ and can label tones without reference to other tones. I wonder if some people have ‘perfect pitch’ with regard to this sort of value judgements.

The Predictably Irrational website is also very good, where Ariely has a regularly updated blog and has created free video summaries of each of the chapters.

All come highly recommended.

Link to AITM interview with Dan Ariely.
Link to Predictably Irrational website.

The New York Times has a thought-provoking article about self-handicapping – the attempt to actually make yourself worse at something. The idea is that if a bad performance is expected, some people actively try and handicap themselves before hand, for example by not practising or by getting drunk, so they have an excuse already lined up and can preserve their self-esteem when they don’t do very well.

I’m sure we’ve all heard about this sort of behaviour discussed anecdotally, but I didn’t realise it’s actually been quite well researched by psychologists since the late 1970s.

Some snippets from the article:

Psychologists have studied this sort of behavior since at least 1978, when Steven Berglas and Edward E. Jones used the phrase ‚Äúself-handicapping‚Äù to describe students in a study who chose to take a drug that they were told would inhibit their performance on an exam (the drug was actually inert)…

Yet given the opportunity, and a good reason, most people will claim some handicap. In a paper [pdf] published last summer, Sean McCrea, a psychologist at the University of Konstanz in Germany, described experiments in which he manipulated participants’ scores on a variety of intelligence tests. In some, the subjects could choose to prepare before taking the test or could join the “no practice” group.

Sure enough, Dr. McCrea found that those told they got bad scores blamed a lack of practice, if they could, and that citing this handicap cushioned the blow to their self-confidence…

As a short-term strategy, self-handicapping is often no more than an exercise in self-delusion. Studies of college students have found that habitual handicappers — who skip a lot of classes; who miss deadlines; who don’t buy the textbook — tend to rate themselves in the top 10 percent of the class, though their grades slouch between C and D.

I wonder how this interacts with the effects of different types of praise and beliefs about intelligence, studied by psychologist Carol Dweck.

She has found that praising a child’s effort on a task (“you’ve worked really hard!”) has a motivating effect, whereas praising the child by attributing their success to a character trait (“you’re really clever”) caused them to become to be more distressed when they encounter failure and lead them to chose easier tasks afterwards.

Her work suggests this is because a belief that intelligence is flexible and effort-related, rather than a fixed character trait, actually makes us more motivated and helps us perform better as we don’t feel we are less intelligent if we fail.

It reflects a likely interaction between performance and self-esteem, mediated by beliefs about competence and I wonder whether self-handicapping is way some people develop to manage this interaction.

Anyway, enough speculation, but I recommend reading the article as it highlights an area that I wasn’t aware of and has many intriguing possibilities.

Link to NYT article on self-handicapping.

The latest edition of The Economist has an interesting article about whether our ability to count and estimate quantity is an innate ability that we have from birth.

The article covers studies on babies, people who speak languages that only have number words for “one”, “two”, “few” and “many”, people who have never developed certain maths skills and others who have lost specific number abilities after brain injury:

Lisa Cipolotti, a neuropsychologist, studied a Signora Gaddi, who used to run a hotel and keep its accounts. After a stroke she could find the number of things in a small group only by counting—when asked how many arms a crucifix had, she got Dr Cipolotti to hold out her arms so she could count them. Signora Gaddi’s problems seemed to affect only numbers. She could still read, speak and reason, remember historical and geographical facts, and order objects by their physical size.

In fact, Signora Gaddi’s difficulties went even deeper than Charles’s. The stroke which damaged her innate understanding of small numbers also robbed her of the entire numerical edifice built on that foundation. For her, numbers stopped at four. When asked to count up from one, she got to four and no further. If there were more than four dots on a page she could not count them. She could not say how old she was or how many days were in a week, or even tell the time.

Link to Economist article ‘Easy as 1, 2, 3’.

A new study just published in PLoS One reports that simply using technical-sounding labels for newly popularised medical conditions changes our understanding of the condition itself, leading us to think it is more serious and more less common.

The study is interesting as it speaks to the debate about disease mongering – the over-medicalising of problems that were previously considered unfortunate but normal parts of life.

The research team, led by psychologist Meredith Young, gave 16 descriptions of medical conditions to two groups of participants.

Eight were conditions that were previously not considered medical disorders but have been ‘medicalised’ in the last 10 years with new technical sounding names common in press reports. For example, impotence is now commonly described ‘erectile dysfunction’ while baldness has been labelled ‘androgenic alopecia’.

The other eight conditions were established medical disorders that have medical and everyday names that are both widely used in the popular press, such as stroke (cerebrovascular accident) and heart attack (myocardial infarction).

One group of participants was given descriptions of the conditions with the common names, and the other group was given the technical names, and each were asked to rate how serious it was, how prevalent it was and whether they thought it was a real disease or not.

For the recently medicalised conditions, the technical label led people to rate it as more serious, more less common and more likely to be a real disease.

For the established conditions, the technical name didn’t effect how the condition was perceived.

Simply giving a condition a technical label seems to change our understanding of the condition itself, making it seem more of a risk and more medically significant.

These findings follow-on from another interesting study from Young, where she found that diseases are thought to be more common and serious the more they’re mentioned in the media.

Participants considered diseases that occur frequently in the media to be more serious, and have higher disease status than those that infrequently occur in the media, even when the low media frequency conditions were considered objectively ‘worse’ by a separate group of participants.

We now know that our beliefs about disease and understanding of illness has a significant effect, not only how we cope with the experience, but how the disease takes its course.

Pharmaceutical companies often promote the benefits of their product, but they also regularly attempt to change our understanding of the problem itself, so the use of their medication seems the most sensible option.

However, there are many other players in the public discussion of illness and certain ideas about causes, symptoms and treatments are often pushed by people because it fits in with other agendas they have.

This is particularly relevant for scientific theories and it is no accident that many of the most significant public medical debates in recent years have been over the acceptance of certain explanations – such as the role of the MMR vaccine in autism, the role of neurotransmitters in mental illness, the role of genetics in obesity.

There is no explanation of illness independent of culture and an understanding of how popular ideas influence our personal medical beliefs is an essential part of understanding medicine itself.

Link to study on medical language and perception of illness.
Link to study on press reporting and perception of illness.

Full disclosure: I’m an unpaid member of the PLoS One editorial board.

Ars Technica has a fantastic article on a recent study that found that numerical specifications in adverts have a huge effect on our choices, even when they’re meaningless.

The numbers can be ratings, technical details, supposed representations of quality – it doesn’t seem to matter. In general, bigger is better and the study found that we tend to be swayed by the numbers even when it directly contradicts our experience.

The first test involved megapixels. The authors took a single image, and used Photoshop to create a sharper version, and one with more vivid colors; they told the students that the two versions came from different cameras. When told nothing about the cameras, about 25 percent of the students chose the one that had made the sharper image. But providing a specification reversed that. When told that the other model captured more pixels using a figure based on the diagonal of the sensor, more than half now picked it. When it comes to specs, bigger is better, too, even if the underlying property is the same. Given the value in terms of the total number of pixels captured, the preference for the supposedly high-resolution camera shot up to 75 percent.

The researchers thought this might be a problem with the fact that not everyone is technically minded, so they tried various other experiments with everything from scented oil to ice-cream – all with the same effect.

To quote the researchers “even when consumers can directly experience the relevant products and the specifications carry little or no new information, their preference is still influenced by specifications, including specifications that are self-generated and by definition spurious and specifications that the respondents themselves deem uninformative.”

Link to Ars Technica write-up of study.
Link to study paper.
Link to DOI.

The Boston Globe has an interesting piece on differing decision-making styles and how cognitive science is increasingly recognising the role of emotion in making choices.

It’s shoehorned into a slightly dubious Obama vs McCain premise, but it covers the important relationship between more conscious reflective forms of problem analysis, and more intuitive forms of approach.

Some of the most interesting research in this area has looked at how these systems interfere with each other.

One of my favourite studies used the Iowa Gambling Task, a card game where participants pick from four decks of cards that can either give them wins and losses. There are various version but a common variant is where two decks give a slight overall gain, while the other two give a slight overall loss.

It’s really hard to work out rationally, because there are just too many numbers to keep in your head, but after a while people tend to get an intuitive grasp of which are the best decks to stick with.

One particular study [full text], led by psychologist Cathryn Evans, found that people with a university education actually did worse on this task than people without one, presumably because they tended to over-apply futile rationalist strategies.

In terms of discussing the problem and ways of tackling it, a classic study by Jonathan Schooler found that getting people to talking about their problem-solving strategy actually made people worse at solving problems, particularly for ‘insight problems‘ where the solution lies in your ability to reframe the whole scenario – often in a counter-intuitive way.

Of course, some problems need a measured, thoughtful, analytical approach, whereas in some situations this interferes with the outcome. However, these are largely findings from lab tasks designed to isolate these types of problems whereas in the real world, problems come as a chaotic mix of both elements.

Knowing which strategy to apply is key, but then again, solving this problem is often equally as complex as solving the problem itself.

Link to Boston Globe article ‘The next decider’.

The ability to intuitively estimate the number objects you can see is known as automatic number sense and has been widely studied in the scientific literature, but is usually thought to be separate from the formal and precise maths abilities we learn at school.

A new study just published online in Nature suggests that these abilities are more intertwined than we might think, as the better the number sense of 14 year-olds, the better their formal maths ability.

The researchers, led by psychologist Justin Halberda, flashed up a series of dot patterns to a group of 14 year-old students. Just like the one in the picture.

The kids were asked to indicate the ratio of blue and yellow dots but because the patterns flashed up so quickly, for only one fifth of a second, the kids didn’t have time to count them. They had to rely on a guestimate – their number sense – to give their answer.

After a whole set of these, the researchers calculated each kid’s accuracy, to give a measure of their overall number sense ability.

This in itself isn’t particularly interesting, as number sense has been widely tested and researched in the scientific literature. However, in the past, it’s often been considered a fuzzy, perhaps more ‘primitive’, ability unrelated to formal maths skills.

Owing to the fact that the researchers had access to the children’s maths achievement test scores, all the way back to kindergarten, they tested whether number sense and maths skills were related.

It turns out they were, and automatic number sense accounted for almost a third of the scores on formal maths tests.

This was even after controlling for the fact that some children were generally brighter or quicker than others.

What is not clear is whether just being better at maths means you develop a better number sense, or whether a better number sense encourages better maths skills.

The fact that they are related at all is interesting, however, as it suggest that intuition plays a part in the practice of mathematics – the most logical of pursuits.

Link to scientific paper.
Link to PubMed entry.
Link to write-up from ScienceNow.

Scientific American’s Mind Matters blog covers an interesting study that found that altering people’s belief in free will also altered the likelihood of participants being dishonest in a test of mental ability.

To achieve this, the study used part of Francis Crick’s book The Astonishing Hypothesis that argues against the everyday concept of free will on the basis of neurobiology.

Half of the participants got a passage saying that there is no such thing as free will. The passage begins as follows: “‘You,’ your joys and your sorrows, your memories and your ambitions, your sense of personal identity and free will, are in fact no more than the behavior of a vast assembly of nerve cells and their associated molecules. Who you are is nothing but a pack of neurons.”

The passage then goes on to talk about the neural basis of decisions and claims that ‚Äú…although we appear to have free will, in fact, our choices have already been predetermined for us and we cannot change that.‚Äù The other participants got a passage that was similarly scientific-sounding, but it was about the importance of studying consciousness, with no mention of free will.

After reading the passages, all participants completed a survey on their belief in free will. Then comes the inspired part of the experiment. Participants were told to complete 20 arithmetic problems that would appear on the computer screen. But they were also told that when the question appeared, they needed to press the space bar, otherwise a computer glitch would make the answer appear on the screen, too. The participants were told that no one would know whether they pushed the space bar, but they were asked not to cheat.

The results were clear: those who read the anti-free will text cheated more often! (That is, they pressed the space bar less often than the other participants.) Moreover, the researchers found that the amount a participant cheated correlated with the extent to which they rejected free will in their survey responses.

I wonder how specific this is to a general belief in us lacking free will, or whether it’s more specifically to do with a similar belief but which is particularly tied up with the mechanistic concept that Crick discusses – i.e. we’re all just the function of lots of little parts.

The reason I’m wondering this is because the twelve-step approach to addiction recovery has two free-will reducing principles at its core – namely an admission that you are not in control of your addiction and the belief that you have to give yourself up to a ‘higher power’.

The Mind Matters article goes on to discuss the various interpretations of the study and how it fits with our understanding of the philosophy of free will.

Link to Mind Matters on ‘Free Will vs. the Programmed Brain’ (via fc).
pdf of full text of study.

The BPS Research Digest covers a recent study finding that volunteers are actually more committed than paid staff in an organisation, in line with studies showing that payment tends to reduce people’s productivity and enjoyment for the same work compared to when it’s done for free.

A recent study published in the Quarterly Journal of Economics tested this by asking students to complete ‘IQ test’ style questions for varying amounts of money, or by ‘incentivising’ some students on a charity collection day while others collected for free.

In this paper we have provided quantitatively precise evidence, in a controlled environment, of the effect of the introduction of monetary compensation on performance, which includes a precise comparison of the cases in which the reward was given in different quantities or not given at all. The result has been that the usual prediction of higher performance with higher compensation, when one is offered, has been confirmed: but the performance may be lower because of the introduction of the compensation.

In other words, those who were paid more worked harder than those who were paid less, but the hardest work was done by those not paid anything at all.

Link to BPSRD on the commitment of volunteers.
Link to summary of payment and productivity paper.
pdf of full-text.

Frontal Cortex has an excellent post on the near futility of election coverage and why people tend to vote with what they feel, rather than what they know.

The piece reviews a whole range of studies that have highlighted possible non-issue influences on people’s voting preferences, from the weather to the facial expressions of news presenters.

One other line of research has found that facial structure can predict leadership, allowing people to reliably pick out business leaders or political winners just from a photo of their face.

Advertisers have long known that marketing products on the basis of facts is a lot less effective than marketing on the basis of appeals to emotion, desire and self-image.

While this is often labelled ‘sex sells’, ‘you-can-be-sexy sells’ is just as widely used.

Traditionally, this avenue has not been open to political candidates since it leaves the candidate open to the emotional counter-attack of accusations of impropriety.

After seeing the popularity of the ‘Obama Girl’ video, it struck me that the internet opens up this avenue, as supporters not officially associated with a candidate can now make their own wide-coverage sex sells promotions without ‘sullying’ the name of the official party machine.

As Frontal Cortex notes:

The problem, as political scientist Larry Bartels notes, is that people aren’t rational: we’re rationalizers. Our brain prefers a certain candidate or party for a really complicated set of subterranean reasons and then, after the preference has been unconsciously established, we invent rational sounding reasons to justify our preferences.

Link to Frontal Cortex on ‘Rational Voters?’.

The Economist has a great article taking a wide-angle view of neuroeconomics, asking whether it actually contributes anything useful to our understanding of economic systems or whether its just a personal psychology of gains and losses that won’t actually scale.

The fiercest attack on neuroeconomics, and indeed behavioural economics, has come from two economists at Princeton University, Faruk Gul and Wolfgang Pesendorfer. In an article in 2005, “The Case for Mindless Economics” [pdf], they argued that neuroscience could not transform economics because what goes on inside the brain is irrelevant to the discipline. What matters are the decisions people take—in the jargon, their “revealed preferences”—not the process by which they reach them. For the purposes of understanding how society copes with the consequences of those decisions, the assumption of rational utility-maximisation works just fine.

But today’s neuroeconomists are not the first dismal scientists to dream of peering inside the human brain. In 1881, a few years after William Jevons argued that the functioning of the brain’s black box would not be known, Francis Edgeworth proposed the creation of a “hedonimeter”, which would measure the utility that each individual gained from his decisions. “From moment to moment the hedonimeter varies; the delicate index now flickering with the flutter of the passions, now steadied by intellectual activity, low sunk whole hours in the neighbourhood of zero, or momentarily springing up towards infinity,” he wrote, poetically for an economist.

Part of the scepticism seems to originate from more general reservations about the results of brain scanning studies being over-interpreted, echoing wider concerns in cognitive neuroscience.

What’s interesting though is that the article mentions that neuroeconomics researchers are turning to transcranial magnetic stimulation (TMS) – a technique that alters brain function for a few hundred milliseconds while people are actively completing tasks.

Because TMS alters brain function, it’s not just showing you a correlation like brain scans do. If task performance changes when you’ve altered that brain area you can infer that the particularly part of the cortex you’ve targeted is causally involved in the psychology of the task.

Along these lines, one recent high-profile study [pdf] managed to alter participants’ fairness behaviour in the Ultimatum Game (a common experimental task) when the function of the upper outside surface of the right frontal lobe was disrupted.

Link to The Economist article ‘Do economists need brains?’.
pdf of ‘The Case for Mindless Economics’.
pdf of TMS study on fairness in the Ultimatum Game.

There have been some excellent articles recently on the psychology of time but one of the most fascinating is from Developing Intelligence who look at a new study that suggests our concept of time becomes nonlinear as we look into the future – in other words, not all futures are equal.

The research, led by psychologist Gal Zauberman, riffs on an effect called ‘hyperbolic discounting‘, where immediate rewards seem more valuable than rewards in the future.

Studies have offered people, for example, ¬£5 now, or more money in the future. Despite the fact that in economic terms they’re better off waiting even for a small amount more, people tend to want considerably more money in the future to make the wait ‘worth it’.

As the DevIntell article notes, this has largely been explained by impulsivity in the past, but a new study considers a radical alternative.

What if the effect is not because we’re impulsive, but because our concept of time is non-linear? In other words, we are reasoning rationally but not on the basis of how much additional time there actually is, but how much longer the wait seems.

These are quite different concepts – for example, we know logically that waiting four weeks is exactly four times as long as waiting a week, but it might not feel exactly four times as bad.

The study asked participants how much extra they’d have to be paid to receive a $75 gift voucher, either in 3 months, 1 year or 3 years. They also had to mark a line to indicate how long each wait seemed, from ‘Very Short’ at one end to ‘Very Long’ at the other.

When compared against the actual time, participants seemed to show hyperbolic discounting, but when compared against the subjective judgement the discounting effect disappeared.

The study goes on to test the effect in different ways, but also added another intriguing angle – when participants were asked to estimate the duration of how long various activities would take, essentially better calibrating their subjective time with actual time, the discounting effect was reduced.

I also really recommend another recent DevIntell post on time perception, discussing how cognitive science theories are attempting to explain how we can perceive something that doesn’t have any ‘sensation’ attached to it.

Any if you’re still hungry for more time, science writer Carl Zimmer has an article in Discover Magazine about how the brain keeps track of time.

Link to DevIntell on distortions in future time perception.
pdf of full-text of study.
Link to DevIntell on time perception and time ‘sensation’.
Link to Carl Zimmer’s article on neuroscience of time.

The Economist has a short but sweet article on a new study that has found that asking the same person to make two guesses and averaging the answer is more accurate than any one guess alone, with more time between guesses improving accuracy.

The study is apparently by psychologists Hal Pashler and Ed Vul and has just been published in Psychological Science, but unfortunately the journal website is down at the moment, but I shall link to the original study when it reappears.

According to The Economist though, here’s the punchline:

The two researchers asked 428 people eight questions drawn from the “CIA World Factbook”: for example, “What percentage of the world’s airports are in the USA?” Half the participants were unexpectedly asked to make a second, different guess immediately after they completed the initial questionnaire. The other half were asked to make a second guess three weeks later.

Dr Vul and Dr Pashler found that in both circumstances the average of the two guesses was better than either guess on its own. They also noticed that the interval between the first and second guesses determined how accurate that average was. Second guesses made immediately improved accuracy by an average of 6.5%; those made after three weeks improved the accuracy by 16%.

Link to Economist article ‘The crowd within’.

The Pirah√£ are a tribe in the Brazilian Amazon who apparently don’t have words for specific numbers. A recent study reported by Science News suggests that despite this, the Pirah√£ people can do numerical tasks, challenging the idea that we need number words to think about and recognize exact quantities.

The study was led by psycholinguist Michael Franks who was interested in previous reports that the Pirah√£ only have words for ‘one’, ‘two’ and ‘many’.

Previous researchers had put a single object on a table, asked a Pirah√£ participant “How much is this?”, added another, asked again and so on, while responses were recorded when different words were used for different quantities.

Frank did the same, but also counted down, starting with a large number of objects and taking one away each time.

He got different answers for the same number of objects and it transpired that the words didn’t mean ‘one’, ‘two’ and ‘many’, as previously thought, but ‘few’, ‘some’ and ‘more’.

In fact, the researchers noted that the Pirah√£ have no linguistic method whatsoever for expressing exact quantity, not even ‘one’.

In a subsequent part, the researchers asked the Pirah√£ participants to do several matching tasks. Some just involved the researchers lining up several objects and asking the participants to match the quantity with a different type of object, with some variation for position and grouping.

Other tasks involved the researchers counting out objects and then hiding them, or counting them into a can.

The Pirah√£ were easily able to do the more straightforward matching tasks, but as soon as they needed to transform the number of items across position or after a delay, they started making errors.

The researchers argue that this suggests we don’t need number words to think about quantity, but they are useful tools to augment our memory.

In other words, numbers are a culturally developed cognitive technology allowing us to remember and compare information about quantity over time and across situations.

Link to abstract of scientific study.
Link to Science News article ‘Numbers beyond words’.

This week’s Bad Science rounds-up several intriguing studies that have found that money does more than make the world go round, it changes how we think, feel and perceive.

The piece looks at several studies where participants paid more, or thought they were getting something of a ‘higher value’, even though there was no actual difference in what they received.

In each case, the ‘higher value’ items things had more of an impact. One study is particularly impressive:

A study published in the Journal of the American Medical Association in March subjected 82 healthy subjects to painful electric shocks, offering them pain relief in the form of a pill which was described as being similar to the opiate codeine, but with a faster onset, in a lengthy and authoritative leaflet. In fact it was just a placebo, a pill with no medicine, a sugar pill, like a homeopathy pill. The pain relief was significantly stronger when subjects were told the tablet cost $2.50 than when they were told it cost 10c.

Link to Badscience on money and psychology.

Hot on the heels of a study that found that simply describing a wine as more expensive made it taste better comes the discovery that the same advice is more likely to be followed if it costs more.

The study was led by organisational psychologist Francesca Gino and is covered by the BPS Research Digest:

Dozens of students were asked questions about American history and received small cash prizes for correct answers. The students were either given the option of receiving advice on the correct answers, or advice was imposed on them. Sometimes this advice was free; other times it was paid for out of the students’ winnings.

Crucially, the advice always came from the same source – in the form of the answer that a student from a pilot session had given to the same question – so the quality of advice was held constant regardless of whether it was free or paid for.

Throughout the study, the participants took more account of advice they had paid for than advice they were given free, even though it was made clear to them that the advice was of the same quality. A final study showed the students took even more account of advice if it was made more expensive.

The full text of the study is freely available online as a pdf, although if you’re not convinced of the findings I’m sure Dr Gino would be happy to supply an additional copy for a small fee.

Link to BPSRD on the behavioural value of expensive advice.