Mental Noise: Is It a Flaw or a Feature?
The search for a quieter and more boring mind is on.
Posted June 10, 2021 | Reviewed by Jessica Schrader
- The book "Thinking, Fast and Slow" reviews psychological heuristics and the biases they produce.
- Noise is a feature of human judgment, and reducing noise is mostly beneficial.
- People may disagree on whether it would be best to live in a world where judgmental noise has been eliminated.
When Nobel Laureate Daniel Kahneman publishes a book, it is guaranteed to be a sensation. In Thinking, Fast and Slow, Kahneman (2011) reviewed decades of work on psychological heuristics and the biases they produce. Once biases are explained, what is left is random error, or Noise. This too, Kahneman, Sibony, and Sunstein (2021; hereafter KSS) now assert, is also a flaw in human judgment. The reader is on notice. If you think you have done the hard work debiasing yourself, think again. Chances are your judgments betray a trembling mind. Your judgments are unreliable, and being unreliable, they also lack accuracy. With the help of algorithms, bureaucratized rules, and simple repetition and aggregation, noise can be kept at bay and the world can become a better place. Writing for the book’s dust jacket, Jonathan Haidt exults that implementing KSS’s “advice would give us more profitable businesses, healthier citizens, a fairer legal system, and happier lives.” What’s not to like?
My goal here is not to offer a review of the book. Reviews of it are being published almost daily. KSS themselves might advise not to read just one of them, for it will contain noise, but to read many, and then average their judgments. To single out the review by Criado Perez (2021) in The Guardian because it resonates with me would be to invite error or bias or both. I will pursue a narrower goal by looking for exceptions to the hard claim that noise arises only from our flawed minds, that noise is always bad, and that it must be eliminated.
But do not let me suggest that KSS have no point. Their review of the psychometrics of measurement and judgment is clear and useful. It starts with the metaphor of mind as a measurement instrument. If the mind is akin to a collection of scales or thermostats, its judgments can be modeled as analogues to the outputs or such devices. Judgments, like all measurements, are variable, and this variation can be decomposed into a true portion, systematic distortions (bias—there can be several), and noise. In contrast to imperfect measurement instruments, mathematical operations of the non-probabilistic kind are not noisy. Given the same input, an equation always yields the same output. Working hard to dim the mental noise, KSS raise the impression that the ideal mind is mathematical.
Focusing on the domains KSS have selected for exploration, one wonders how human judges have gotten away with such noisy judgments for so long. Noisy business decisions (e.g., in- or divestments) are costly (Sibony), noisy judicial judgments (e.g., sentencing) are unfair (Sunstein), and noisy everyday judgments are irrational (Kahneman). Clear rules and algorithms do better, as do repeated judgments bundled by aggregation. When algorithms and other denoising procedures have done their job, the mind has been quieted, and it seems that the noise that was initially heard was the mind’s fault, as the book’s byline suggests.
We can challenge the unquestioned attribution of the flaw to the mind. Perhaps part of the blame lies with the stimulus. Some stimuli are ambiguous or visually degraded so that they invite or attract variable judgments from the perceiver. When we judge a room’s temperature without looking at a thermostat, our judgments may contain bias (by either being too high or too low on average) and noise (as one estimate of the temperature differs from the next). The variability of judgment is a function of both the perceiver’s imperfect sensitivity and the stimulus’s imperfect clarity. To place the blame squarely on one or the other obscures this interdependence.
In the case of room temperature, there is a true state of nature, which can be assessed with an instrument of high fidelity, whose judgment should displace our intuition. What if there is no true state? Consider Leonardo’s Mona Lisa. Her expression is ambiguous. Is there a smile playing around her lips, and if there is, how subtle is it? Not only do art enthusiasts and ordinary visitors of the Louvre fail to reach perfect consensus, so may individual perceivers have different impressions on different occasions. Since the Mona Lisa is what she is, a static stimulus, it may be tempting to look for the source of this variability in the perceivers’ minds, but it is a property of the painting to bring this about. At a perceptual level once removed, one may say that the Mona Lisa would be a less famous painting were it not so.
The Necker Cube is a more radical example of a stimulus provoking conflicting perceptions. As a stimulus, the Necker is underdetermined (Attneave, 1971). Indeed, if we all agreed that we are looking at the insides and not the outsides of the cube, we should worry that there is something wrong with our perceptual apparatus. Being able to generate variable, even conflicting, perceptions may rather be a design feature of our minds than a flaw. Granted, KSS acknowledge the issues of objective ignorance, that is, limits to perception, and the salutary roles of creativity and dissent, but they fail to accord them the full respect they deserve. Their treatment of the matter is of the gotcha! variety. Look, they seem to be saying, here's another immense type of human fallibility, and, what is more, the failure of psychological science to not having noticed it until we did.
Important as KSS’s insights and demonstrations are, it is useful to consider their limits and blind spots. Ask yourself, would you want to live in a world where judgmental noise has been eliminated, not only in the domains where it is clearly costly or unjust, but everywhere? At the limit, do we wish to let the algorithms do all the thinking, leaving us free to go to the movies? Is this a u- or a dystopia? Reasonable people may disagree on this, and their perceptions should not be averaged—in my opinion.
On a lighter note, the original title of this post was Kahneman's Noise, which I intended to be an ambiguous stimulus, kind of like Mona L. There are (at least) 2 meanings, and they can't be averaged.
Attneave, F. (1971). Multistability in perception. Scientific American, 225, 63–71.
Criado Perez, C. (2021, June 3). Noise’by Daniel Kahneman, Olivier Sibony and Cass Sunstein review – the price of poor judgment. https://www.theguardian.com/books/2021/jun/03/noise-by-daniel-kahneman-…; retrieved on Jun 10, 2021.
Kahneman, D. (2011). Thinking, fast and slow. Farrar, Straus & Giroux.
Kahneman, D., Sibony, O., & Sunstein, C. (2021). Noise: A flaw in human judgment. Little, Brown Spark.