Wednesday, January 29, 2020

Not born yesterday

I am looking forward to this, Not Born Yesterday: The Science of Who We Trust and What We Believe by Hugo Mercier.

The last twenty, maybe thirty years has seen a barrage of stories and research about the failure of the human epistemic system. We are unreliable witnesses, we are not logical, we are not open to new data, we are biased, we are subjective, we are overconfident in our own capabilities, we lack self-control, etc. While all true to some degree in some instances, there is kind of a problem with this line of argument if you have any confidence in darwinian evolution.

Whatever is going on in the grey matter, it clearly is effective, perhaps even optimally effective. There are more people living better lives than at any point in human history. We may not be perfect thinkers but something is working.

While there are elements of truth in much of the above criticism there are a couple of errors in the analysis.

While the criticisms always encompass some valid issue, the extent and consequence of the problem are almost always overstated. In many instances, subsequent research reveals that the analysis itself was wrong in the first place.

Perhaps the biggest flaw in the massive industry of proving human thinking fallibility is that elements are examined in isolation and without context. If people operate with some simple heuristic which has a high error rate, the issue is not that there is an error rate. The issue is whether there is a better heuristic with a lower error rate.

We encounter all sorts of people and never have enough time to really get to know them well enough to make the types of judgments we need to make in the short time-frames we have. All decision-making entails expenditure of time and energy and we never have enough time and never enough energy in an immensely complex world with inherent uncertainty to make the very best decisions possible. Instead, we find heuristics and cognitive shortcuts which are on balance economically beneficial.

When we rely on heuristics, we are playing the odds. We won't get every decision right but we may get them mostly right given the time and energy constraints within which we operate.

A made up example. Say I operate with the heuristic that I should place more confidence in tall people. It will be easy to demonstrate that this is an unreliable heuristic. There will be plenty of instances where trusting a particular tall person is a bad idea. But given that it seems true that tall people disproportionately end up in leadership positions, perhaps over multiple instances and over time, in aggregate, having confidence in tall people might be an efficient heuristic even though it is not a uniformly accurate one.

The question is not whether placing confidence in tall people is a uniformly and completely accurate strategy. The question is whether there is a superior heuristic that yields better results.

Most research seeks lab room perfection - i.e. they examine human cognition against a benchmark of perfection. In the real world of constraints, the benchmark is not perfection but optimization. What habits yield the greatest long run benefit at the lowest short term cost.

My aversion for the conventional academic narrative about the flaws of human epistemology goes beyond the test of demonstrated evolutionary fitness.

Too often, it appears to me that much of the narrative of flawed human epistemology is a means towards an ends. Very bright people (only some of whom are accomplished) trying to demonstrate why we would be all better off by leaving the hard decision-making to our cognitive betters. This implicit assumption is part of my aversion to Nudge by Cass Sunstein. It presumes, despite their record of failure, that centralized decision-making by experts is superior to dispersed decision-making by all citizens. And using evidence of epistemic shortcomings to justify that centralization of decision-making.

Give me Darwin (evolution) and Ferguson (emergent order) over Marx (centralized decision-making by the vanguard).

Looks like Mercier may be making the arguments I so rarely see made. The optimally effective decision-making under time and resources constraints of course looks radically different from perfect decision-making without any time and resource constraints.

From the blurb:
Why people are not as gullible as we think

Not Born Yesterday explains how we decide who we can trust and what we should believe--and argues that we're pretty good at making these decisions. In this lively and provocative book, Hugo Mercier demonstrates how virtually all attempts at mass persuasion--whether by religious leaders, politicians, or advertisers--fail miserably. Drawing on recent findings from political science and other fields ranging from history to anthropology, Mercier shows that the narrative of widespread gullibility, in which a credulous public is easily misled by demagogues and charlatans, is simply wrong.

Why is mass persuasion so difficult? Mercier uses the latest findings from experimental psychology to show how each of us is endowed with sophisticated cognitive mechanisms of open vigilance. Computing a variety of cues, these mechanisms enable us to be on guard against harmful beliefs, while being open enough to change our minds when presented with the right evidence. Even failures--when we accept false confessions, spread wild rumors, or fall for quack medicine--are better explained as bugs in otherwise well-functioning cognitive mechanisms than as symptoms of general gullibility.

UPDATE: Not two hours after I posted this, the perfect example of tactically optimal heuristic decision-making under uncertainty and constraints pops up on Twitter.



No comments:

Post a Comment