Stumbling and Mumbling

Author: chris dillow   |   Latest post: Tue, 10 Dec 2019, 1:48 PM


Seeing what we expect

Author:   |    Publish date:

Read the words in the triangles. Wrong1zqipfewrtn3w_small

Now read them again.

If you are like most people unprepared for the task, you'll have read even these very simple sentences wrongly. You'll have committed the mistake of seeing what you expect rather than what is there; you are so used to seeing such familiar phrases that you expect to see them even when what is actually there is a slight variation on them.

This is why it is impossible for most of us to proofread our own writing.

Not that this is always a handicap. Filling in gaps and overlooking small errors is why we can make sense of what people are saying when we can't easily hear them because of background noise. And it's why you all understand sentences such as "Boris Johnson is a c*** who is talking s***".

Sometimes, though, this can lead us astray. It is, I suspect, one reason (of several) for the momentum effect in stock markets - the tendency for winning stocks to carry on rising and losers to continue falling. Investors who have bought a stock and then see it suffer some bad news cleave too strongly to their prior belief that it was a good stock and so don't sell it sufficiently strongly. This causes the share to be over-priced and so to drift down gradually later.

This is Bayesian conservatism - the tendency to be too conservative in updating our beliefs when we see disconfirming evidence; we see what we expect, rather than what is there. It is often underpinned by two other common biases. One is wishful thinking: we want to believe that we were right in the first place, and the wish is father to the belief. And the other is overconfidence: we exaggerate our intellectual ability and so overestimate the extent to which we are right.

Of course, this is not the only deviation we make from full rationality. Just as it's possible to fall off a bike to the left as well as to the right, so we can make the opposite mistake of underweighting prior information. This is called base rate neglect. It is, I suspect, one reason why bosses and investors overestimate the benefits of mergers: they pay too much attention to claims about synergy, and not enough to historic evidence that such mergers often fail.

We have, though, seen two examples of the mistake of seeing what we expect recently. One is David Ornstein's report that Arsenal's bosses are "100% behind Unai Emery." If they are sincere (and they might be), they are seeing what they expect - that Emery is the right man - rather than what the evidence shows*.

The other was Jonathan Freedland's mistaken claim that Labour has shortlisted an anti-semite as candidate for Birmingham Hall Green. I suspect that Mr Freedland's willingness to think of Labour as an anti-semitic party led him to infer too quickly that the Majid Mahmood they had shortlisted was the same Majid Mahmood who was an anti-semite when in fact they are different men. Had your prior been that Labour could not possibly select an anti-semite, you would have reacted to the facts "Labour selects Majid Mahmood" and "Majid Mahmood was fined for anti-semitism" by checking whether they were in fact different Mr Mahmoods. Maybe your prior was wrong, maybe not - but it would at least have led you to the facts.

It would be harsh to criticise Mr Freedland for seeing what he expected: this is a mistake we all make. Where he erred was in having a prior belief that so easily led to error in this case.

You might think that episode shows that centrists are not the rational evidence-based pragmatists they claim to be, but are in fact as ideological and irrational as the rest of this. I'm tempted to agree. But then in doing so I might commit exactly the same error - of seeing what we expect.

* There are parallels between those who wanted Wenger sacked and Brexiters: both demanded change without sufficient consideration of the effects of doing so.

Share this

  Be the first to like this.


2107  2164  644  2694 

Top 10 Active Counters
 LLOY 64.33+3.21 
 PREM 0.09-0.008 
 ORCP 1.30-0.075 
 UJO 0.16+0.007 
 BOU 0.024+0.001 
 BARC 182.48+10.62 
 RLD 0.36+0.06 
 VILX 0.464-0.023 
 TW 199.65+25.55 
 EUZ 0.025-0.001 
Partners & Brokers