Showing posts with label confirmation bias. Show all posts
Showing posts with label confirmation bias. Show all posts

Wednesday, May 9, 2007

The White Crow




This post is another response to another discusson on Overcoming Bias.

Robin Hanson is addressing disagreements in the context of the cold fusion controversy.

How can they each reconcile their own view with the fact that smart expert people are on the other side?

The positions of accepting and denying a phenomenon are not symmetric. This was pointed out by William James with his "White Crow" analogy. You only need one white crow to disprove the rule that all crows are black.

In a similar way, a large number of failed cold fusion replications are irrelevant if we can find a single experiment that provides irrefutable data. Of course there is no such thing as a single irrefutable experiment, but in the case of LENR / cold fusion there are a whole host of good experiments demonstrating some kind of anomalous effect.

The principle problem is that people, even very smart and motivated professional scientists, are very much driven by their theories on how the world must be. In fact, professional scientists are probably much more theory-driven than the average person, and therefore more inclined to confirmation bias.

We see a good example of this in Eliezer's remark in his afterlife post that cryonics, actuarial escape velocity, and nanotechnology roads to immortality were preferable to a hypothesized afterlife, because they "put far less of a strain on the Standard Model".

The truth-seeking approach is to consider the full spectrum of available evidence to determine the correctness of our models, instead of using our models to determine what evidence is correct. It is the only way to avoid dogmatism and discover what is real.

Saturday, April 7, 2007

Confirmation Bias (Repost from AMNAP 1.0)



Emory University released a study in January of 2006 examining fMRI images of brain activity in registered Democrats and Republicans who were watching videos of George Bush and John Kerry caught in political untruths. The results of the study were fascinating (emphasis added by me):


When it comes to forming opinions and making judgments on hot political issues, partisans of both parties don't let facts get in the way of their decision-making, according to a new Emory University study. The research sheds light on why staunch Democrats and Republicans can hear the same information, but walk away with opposite conclusions.

The investigators used functional neuroimaging (fMRI) to study a sample of committed Democrats and Republicans during the three months prior to the U.S. Presidential election of 2004. The Democrats and Republicans were given a reasoning task in which they had to evaluate threatening information about their own candidate. During the task, the subjects underwent fMRI to see what parts of their brain were active. What the researchers found was striking.

"We did not see any increased activation of the parts of the brain normally engaged during reasoning," says Drew Westen, director of clinical psychology at Emory who led the study. "What we saw instead was a network of emotion circuits lighting up, including circuits hypothesized to be involved in regulating emotion, and circuits known to be involved in resolving conflicts.". . .

Once partisans had come to completely biased conclusions -- essentially finding ways to ignore information that could not be rationally discounted -- not only did circuits that mediate negative emotions like sadness and disgust turn off, but subjects got a blast of activation in circuits involved in reward -- similar to what addicts receive when they get their fix, Westen explains.

"None of the circuits involved in conscious reasoning were particularly engaged," says Westen. "Essentially, it appears as if partisans twirl the cognitive kaleidoscope until they get the conclusions they want, and then they get massively reinforced for it, with the elimination of negative emotional states and activation of positive ones."

During the study, the partisans were given 18 sets of stimuli, six each regarding President George W. Bush, his challenger, Senator John Kerry, and politically neutral male control figures such as actor Tom Hanks. For each set of stimuli, partisans first read a statement from the target (Bush or Kerry). The first statement was followed by a second statement that documented a clear contradiction between the target's words and deeds, generally suggesting that the candidate was dishonest or pandering.

Next, partisans were asked to consider the discrepancy, and then to rate the extent to which the person's words and deeds were contradictory. Finally, they were presented with an exculpatory statement that might explain away the apparent contradiction, and asked to reconsider and again rate the extent to which the target's words and deeds were contradictory.

Behavioral data showed a pattern of emotionally biased reasoning: partisans denied obvious contradictions for their own candidate that they had no difficulty detecting in the opposing candidate. Importantly, in both their behavioral and neural responses, Republicans and Democrats did not differ in the way they responded to contradictions for the neutral control targets, such as Hanks, but Democrats responded to Kerry as Republicans responded to Bush.

While reasoning about apparent contradictions for their own candidate, partisans showed activations throughout the orbital frontal cortex, indicating emotional processing and presumably emotion regulation strategies. There also were activations in areas of the brain associated with the experience of unpleasant emotions, the processing of emotion and conflict, and judgments of forgiveness and moral accountability.

Notably absent were any increases in activation of the dorsolateral prefrontal cortex, the part of the brain most associated with reasoning (as well as conscious efforts to suppress emotion). The finding suggests that the emotion-driven processes that lead to biased judgments likely occur outside of awareness, and are distinct from normal reasoning processes when emotion is not so heavily engaged, says Westen.

The investigators hypothesize that emotionally biased reasoning leads to the "stamping in" or reinforcement of a defensive belief, associating the participant's "revisionist" account of the data with positive emotion or relief and elimination of distress. "The result is that partisan beliefs are calcified, and the person can learn very little from new data," Westen says.

The study has potentially wide implications, from politics to business, and demonstrates that emotional bias can play a strong role in decision-making, Westen says. "Everyone from executives and judges to scientists and politicians may reason to emotionally biased judgments when they have a vested interest in how to interpret 'the facts,' " Westen says.

Coauthors of the study include Pavel Blagov and Stephan Hamann of the Emory Department of Psychology, and Keith Harenski and Clint Kilts of the Emory Department of Psychiatry and Behavioral Sciences.


It is obvious that these results--if themselves confirmed--could easily have a broad application to any area of strong beliefs and heated debate. Is it true that skeptics of psi phenomena are being irrational about the facts? There is certainly some data pointing in this direction. Of course, those who accept the existence of psi phenomena need to watch for their own tendencies towards confirmation bias, and remember that many psychics and purported phenomena are bogus. This definitely adds support to the charge that "scientific" opposition to psi phenomena is more about sociology than science.