" We do so not because we are interested in mistakes for their own sake, but because correct thinking is the default case, which requires no explanation."
"I am still shocked Kahneman wrote this. Correct thinking is the default case and requires no explanation? What?!"
As you say, heuristic beliefs only become biases when they miss the target. ‘Correct thinking’ here sounds like when the belief is conducive to hitting the target, and therefore ‘requires no explanation’ because no error occurred, so it is unlikely to inspire reflection on the belief that underlies the action. The predictive error that occurs when applying the belief in a situation where it is not conducive to the desired outcome is what highlights and triggers reflection on the underlying belief.
I think rational models are perhaps best understood as schematic training wheels for the development of expert intuition. They provide a tentative framework for the understanding of how systems interact within a given domain, that is then fleshed out and re-organized through failed attempts to apply it to reality, as the unpredicted outcomes defy assimilation and demand the deeper processing necessary for accommodation. The problem with rational models occurs when academics get their ego wrapped up in their pet maps and mistake them for the territory.
I haven't read this book (only about it), but L.A. Paul's Transformative Experience would function as a counterargument to Thaler's quip: it's about how decision theory breaks down in the face of certain kinds of pivotal life decisions that transform you by virtue of undergoing them; a change which by definition you cannot calculate in advance of living through it.
Re. your perplexity in footnote 6 over K's statement that "Correct thinking is the default case and requires no explanation": I actually read this to mean "correct" in its commonsense usage of some fairly obvious local standard or goal that enables us to distinguish error from success (e.g. mathematically correct answer to a bat and ball question); not normative thinking more generally. You don't need a controversial normative theory to talk about errors, only for biases.
Fair point. I don't really understand the naming conventions of comics, but this Reddit post has it as "Spider-Man fact sheet written by Stan Lee and Drawn by Steve Ditko (Spider-Man Annual #1 1964)"
Excellent article! Now I look forward to/dread learning more about non-ergodicity.
You might also be interested in an upcoming article titled "Biases as Errors of Omission." Rather than reclassifying all biases as errors of omission, the move is to show that, over time, the concept of "bias" has skewed toward a rhetorical accusation as if they are errors of commission. Complementary to your point here, we should expect heuristics, by their very definition as simplifications and shortcuts, to risk errors of omission during the process of simplification. We should also expect repeated uses of similar heuristics to produce similar omissions.
So the post will introduce "Lens Neglect" and "Deferential Survivorship Bias" (mirroring Base Rate Neglect and [Differential] Survivorship Bias) as systemic errors of omission that better capture the necessary tradeoffs in affording quality heuristics. They may also better identify points of intervention than memorizing 100+ errors and "just stopping" cold turkey. I am also introducing a concept that mirrors K&T's Attribute Subsitution (~affording rhetorical confidence): "Feature Substitution" — substituting a feature-positive simplification for a feature-negative phenomenon. For example, your point about biases technically not existing could be framed as an attribute substitution in the sense that reifying biases as measurable deviations (from a possibly arbitrary standard) affords rhetorical confidence. Conversely, the technical non-existence of "correlations" could be framed as feature-positive substitutions for conspicuous non-deviation (technically, a feature-negative relation).
Also, in my most recent post, I made a similar point about confirmation bias. First, it's worth pointing out how the rhetorical accusation of "avoiding" contrary information and "seeking" confirming information is an error-of-commission framing of what was originally studied as an error of omission. The rhetorical escalation also tracks the frustrations of social scientists trying to force falsificationism (great at maintaining a high quality information environment) into politics — a notoriously low quality information environment in which journalistic "corroboration" (a kind of verificationism) has evolved. "Confirmation bias" conspicuously tracks deviation from presumptive-falsificationism, even when falsificationism is susceptible to common strategies, such as "flooding the zone."
I have lots of thoughts (don't get me started on utilitarianism...). I'll stick to one: GMOs. The problem with GMOs is not so much health* as it is wealth, and in particular the malicious behavior of large corporations.
My concern is that corporations like Monsanto will use GMO technology to monopolize markets, such as soy, and then use their monopoly power to exploit farmers, especially undereducated and poor ones in the third world. (Monsanto has form for this.) One example that is already out there is GMO soy beans, supposedly a boon to the farmer and the consumer for their greater nutrition.
Except that Monsanto also engineered these patented beans to be self-terminating: farmers cannot, as they have done for 10,000 years**, plant some of this years crop to grow next years crop. Every year, they have to go back to Monsanto for more beans. Or they can grow fewer and worse beans than their neighbors.
Monsanto then compounded this by producing its weedkiller RoundUp, to which its GMO beans have been engineered to be resistant, so you can spray it liberally without concern for whether, for instance, it happens to get blown onto your neighbor's non-resistant crops.
In short, this is the danger of GMO as a technology: not that the foods might be more dangerous than naturally evolved foods; but that monopolistic corporations will use them dominate markets and enrich themselves at the cost of the farmers. Which is exactly the kind of thing agrobusinesses have done in the past.
Of course, there are ways to get the benefits of GMOs without these risks, for example governments could ban terminating seeds. Good luck with that.
By the way, this point rarely gets heard because people who object to GMOs for _any_ reason get told they are being scientifically illiterate and, dare I say it, irrational.
----
*OK, it is actually health too. RoundUp, whose widespread adoption was a side-affect of the adoption of GMO seeds, is now widely regarded as carcinogenic.
Obviously not the focus of this piece, but perhaps as the author I have the duty to respond.
Arguing against specific business practices is fine, and I probably agree with at least some of those arguments. Each of the points you made above is an empirical argument and we could have an actual debate about whether Monsanto engaged in those practices, and the consequences of their actions.
But the "all natural" movement is fundamentally about health, not business practices. And when it comes to health, there isn't really a debate to be had. The argument that GMOs are less healthy doesn't make sense even in theory, and in practice GMO's are typically healthier and contain less pesticide than products labeled as "organic" (in quotes because the term has no scientific definition).
To put it another way, when Green Peace burnt the fields of Golden Rice (a crop designed to cure blindness), it wasn't because they were worried about unethical business practices, but because they have a pre-scientific belief that messing with nature is immoral. And it is in cases like that where I am tempted to use the word "irrational."
But if you simply want to criticize a companies business practices, then that is completely rational.
" We do so not because we are interested in mistakes for their own sake, but because correct thinking is the default case, which requires no explanation."
"I am still shocked Kahneman wrote this. Correct thinking is the default case and requires no explanation? What?!"
As you say, heuristic beliefs only become biases when they miss the target. ‘Correct thinking’ here sounds like when the belief is conducive to hitting the target, and therefore ‘requires no explanation’ because no error occurred, so it is unlikely to inspire reflection on the belief that underlies the action. The predictive error that occurs when applying the belief in a situation where it is not conducive to the desired outcome is what highlights and triggers reflection on the underlying belief.
I think rational models are perhaps best understood as schematic training wheels for the development of expert intuition. They provide a tentative framework for the understanding of how systems interact within a given domain, that is then fleshed out and re-organized through failed attempts to apply it to reality, as the unpredicted outcomes defy assimilation and demand the deeper processing necessary for accommodation. The problem with rational models occurs when academics get their ego wrapped up in their pet maps and mistake them for the territory.
I haven't read this book (only about it), but L.A. Paul's Transformative Experience would function as a counterargument to Thaler's quip: it's about how decision theory breaks down in the face of certain kinds of pivotal life decisions that transform you by virtue of undergoing them; a change which by definition you cannot calculate in advance of living through it.
Re. your perplexity in footnote 6 over K's statement that "Correct thinking is the default case and requires no explanation": I actually read this to mean "correct" in its commonsense usage of some fairly obvious local standard or goal that enables us to distinguish error from success (e.g. mathematically correct answer to a bat and ball question); not normative thinking more generally. You don't need a controversial normative theory to talk about errors, only for biases.
Minor point, but is the comic art maybe S. Ditko or AI? Credit where due.
Fair point. I don't really understand the naming conventions of comics, but this Reddit post has it as "Spider-Man fact sheet written by Stan Lee and Drawn by Steve Ditko (Spider-Man Annual #1 1964)"
https://www.reddit.com/r/Spiderman/comments/s6irvo/spiderman_fact_sheet_written_by_stan_lee_and/
“Inconceivable!”
Excellent article! Now I look forward to/dread learning more about non-ergodicity.
You might also be interested in an upcoming article titled "Biases as Errors of Omission." Rather than reclassifying all biases as errors of omission, the move is to show that, over time, the concept of "bias" has skewed toward a rhetorical accusation as if they are errors of commission. Complementary to your point here, we should expect heuristics, by their very definition as simplifications and shortcuts, to risk errors of omission during the process of simplification. We should also expect repeated uses of similar heuristics to produce similar omissions.
So the post will introduce "Lens Neglect" and "Deferential Survivorship Bias" (mirroring Base Rate Neglect and [Differential] Survivorship Bias) as systemic errors of omission that better capture the necessary tradeoffs in affording quality heuristics. They may also better identify points of intervention than memorizing 100+ errors and "just stopping" cold turkey. I am also introducing a concept that mirrors K&T's Attribute Subsitution (~affording rhetorical confidence): "Feature Substitution" — substituting a feature-positive simplification for a feature-negative phenomenon. For example, your point about biases technically not existing could be framed as an attribute substitution in the sense that reifying biases as measurable deviations (from a possibly arbitrary standard) affords rhetorical confidence. Conversely, the technical non-existence of "correlations" could be framed as feature-positive substitutions for conspicuous non-deviation (technically, a feature-negative relation).
Also, in my most recent post, I made a similar point about confirmation bias. First, it's worth pointing out how the rhetorical accusation of "avoiding" contrary information and "seeking" confirming information is an error-of-commission framing of what was originally studied as an error of omission. The rhetorical escalation also tracks the frustrations of social scientists trying to force falsificationism (great at maintaining a high quality information environment) into politics — a notoriously low quality information environment in which journalistic "corroboration" (a kind of verificationism) has evolved. "Confirmation bias" conspicuously tracks deviation from presumptive-falsificationism, even when falsificationism is susceptible to common strategies, such as "flooding the zone."
Keep'em comin'! Cheers!
I have lots of thoughts (don't get me started on utilitarianism...). I'll stick to one: GMOs. The problem with GMOs is not so much health* as it is wealth, and in particular the malicious behavior of large corporations.
My concern is that corporations like Monsanto will use GMO technology to monopolize markets, such as soy, and then use their monopoly power to exploit farmers, especially undereducated and poor ones in the third world. (Monsanto has form for this.) One example that is already out there is GMO soy beans, supposedly a boon to the farmer and the consumer for their greater nutrition.
Except that Monsanto also engineered these patented beans to be self-terminating: farmers cannot, as they have done for 10,000 years**, plant some of this years crop to grow next years crop. Every year, they have to go back to Monsanto for more beans. Or they can grow fewer and worse beans than their neighbors.
Monsanto then compounded this by producing its weedkiller RoundUp, to which its GMO beans have been engineered to be resistant, so you can spray it liberally without concern for whether, for instance, it happens to get blown onto your neighbor's non-resistant crops.
In short, this is the danger of GMO as a technology: not that the foods might be more dangerous than naturally evolved foods; but that monopolistic corporations will use them dominate markets and enrich themselves at the cost of the farmers. Which is exactly the kind of thing agrobusinesses have done in the past.
Of course, there are ways to get the benefits of GMOs without these risks, for example governments could ban terminating seeds. Good luck with that.
By the way, this point rarely gets heard because people who object to GMOs for _any_ reason get told they are being scientifically illiterate and, dare I say it, irrational.
----
*OK, it is actually health too. RoundUp, whose widespread adoption was a side-affect of the adoption of GMO seeds, is now widely regarded as carcinogenic.
**or however long humans have been planting seeds
Obviously not the focus of this piece, but perhaps as the author I have the duty to respond.
Arguing against specific business practices is fine, and I probably agree with at least some of those arguments. Each of the points you made above is an empirical argument and we could have an actual debate about whether Monsanto engaged in those practices, and the consequences of their actions.
But the "all natural" movement is fundamentally about health, not business practices. And when it comes to health, there isn't really a debate to be had. The argument that GMOs are less healthy doesn't make sense even in theory, and in practice GMO's are typically healthier and contain less pesticide than products labeled as "organic" (in quotes because the term has no scientific definition).
To put it another way, when Green Peace burnt the fields of Golden Rice (a crop designed to cure blindness), it wasn't because they were worried about unethical business practices, but because they have a pre-scientific belief that messing with nature is immoral. And it is in cases like that where I am tempted to use the word "irrational."
But if you simply want to criticize a companies business practices, then that is completely rational.