Thursday, September 17, 2009
How Quantum Probability Theory Could Explain Human Logical Fallacies
A quantum model of reasoning beats its classical counterpart in explaining why humans make errors in judging probabilities.
The conjunction and disjunction fallacies are famous for revealing
the limits of human reasoning about probability.
This can be measured by telling people a short story about a
character and then asking questions about the likelihood of certain
statements about that character. Take a look at this story about
Linda (which I've taken from Wikipedia):
Linda is 31 years old, single, outspoken, and very
bright. She majored in philosophy. As a student, she was deeply
concerned with issues of discrimination and social justice, and also
participated in anti-nuclear demonstrations.
Which is more probable?
Linda is a bank teller.
Linda is a bank teller and is active in the feminist
movement.
It turns out that 85 per cent of people choose the second option. But the probability of two events occurring together (in
conjunction) is always less than or equal to the probability of one
of them alone.
This is the conjunction fallacy (humans show a
similar problem over the probability of one event OR another being
true, called the disjunction fallacy).
The question is how to
explain the problem humans have with this kind of reasoning. Until
now, psychologists have turned to classical probability theory to
study the concept of probability judgement error. This allows
them to build a mathematical model of human reasoning that
allows for errors in judgement.
But Jerome Busemeyer at
Indiana University and buddies have a different take. They say that
quantum probability theory leads to more realistic predictions about
the type of errors humans make.
"Quantum probability
theory is a general and coherent theory based on a set of (von
Neumann) axioms which relax some of the constraints underlying
classic (Kolmogorov) probability theory," say the team.
That's
an interesting insight, to say the least. And if it pans out, it
signals a fundamental shift in thinking about the brain.
What
Busemeyer and co are saying is that the principles of quantum
information processing, including the ideas of superposition and
interference, lead to better models of the way humans make
decisions.
What this idea needs, of course, is some kind
of testable hypothesis that differentiates it from classical models.
The team hint at this when describing how the principle of
superposition applies to thinking about voting habits, when a voter has to choose between two
candidates.
According to classical theory, before the vote is
cast, the voter is in a mixed state. But Busemeyer and co say that
thinking about the voter in a superposition of states is a better
model. That kind of thinking ought to lead to some testable
predictions.
Busemeyer and co are at pains to distance
themselves from research that uses quantum mechanics to model the
brain in an attempt to understand consciousness. and memory. "We
are not following this line," they say. Instead they keep their
work far more abstract.
But inevitably the question will be
asked. If the principles of quantum information processing better
describe the way humans make decisions, what does that imply about the
way the brain works?
There's no telling where this kind of thinking will lead.
Ref: arxiv.org/abs/0909.2789: Quantum Probability Explanations for Probability Judgment 'Errors'
Massachusetts Institute of Technology
Comments
The point is that the sort of system which you use to work out quantum probabilities has similarities with the sort of processes people use to work out certain logical problems, and get them wrong. It doesn't really mean the brain is using quantum physical processes (at least, no more than any other classical physical system) to make decisions. The "superposition of beliefs during uncertainty" is only metaphorically connected to quantum superposition. You could just as well say, the subject "holds the possibility of the truth of two conflicting beliefs during times of uncertainty".
The researchers "are at pains" to emphasise this, as reported. But at the end of the article the journalist gives a hefty hint in the "quantum brain activity" direction anyway.
GreenwichMan
09/18/2009
Posts:1
Judging the probability of some hypothetical occurrence is not something that people are intrinsically good at. On a toy problem, such as the probability of the unknown Mary being a bank teller, one is clutching at straws to come to a conclusion. Even in a real world situation, probability judgments are poor. For the USSR invading Poland example, the 4% guess at a probability -- if also given for invasions of Hungary and Czechoslovakia -- would mean the odds are in favor of an invasion sometime in the next six years (assuming the same likelihood each year and a very small probability of multiple invasions).
It is perfectly reasonable to assume that judgments involving a single conclusion and those involving multiple conclusions are mentally processed differently. If one is given additional information, then one is likely to judge the probabilities differently.
Thus for Mary, the first judgment might be the probability that she would be a stereotypical bank teller, while the second would be whether she would be a feminist bank teller. The odds of the second would be judged higher than the first. And if many feminist bank tellers are not stereotypical, the relative judgments could very well be correct.
In weighing the differences in concluded likelihoods, many respondents may respond by comparing their gut feelings of the probability of the two situations without making the logical comparison between the meanings of the two analyzed states.
The result of having quantum probability reasoning (which deals with both independent and coupled probabilities) more accurately describe human probability judgments than logical reasoning which considers such reasoning mechanisms dependent does not mean that any quantum mechanical explanations are needed. The paper is clear about this, while this post suggests that there may be a link.
dougf
09/22/2009
Posts:1
"Superposition" seems like a good equivalent to positing "I'm having my own little 'run-off' election in my head, and the winner THERE will be the person I vote for" -- which is what most people do, I suspect.
Reframing that as "I'm hosting two totally different mental frameworks, one where I'm planning to vote for THIS guy, and another where I'm planning to vote for THAT guy" also works to approach the "constantly-distracting 'many voices' paradigm".
That's the one where all the internal "independent evaluators" generate a consensus over time, and then the voter quite rightfully can say, just like the bumper sticker indicates, "I just do what the voices tell me..."
flared0ne
09/29/2009
Posts:47
Example: We are detective and we know, killer wears bra and we know, killer was speaking Spanish. Now we have Chinese guy, Spanish woman and British woman and we have to find the murderer between them.
The option, killer is boy isn't symmetrical to option, killer was a women - because WE KNOW, boys doesn't wear bra and option(s) other then man or woman aren't possible here (this is hidden constrain in the task). The assumption, killer was a just a boy or woman has lower probability here, then the assumption, killer was Spanish woman, after then.
In real decisions the number of options is often limited or constrained mutually, so that classical probability theory related to infinite number of possible options doesn't work here.
ZephirAWT
09/30/2009
Posts:158
ZephirAWT
09/30/2009
Posts:158
the linda story .......i think the reason people get it wrong is not because they are stupid or " probability blind " - the way the question is phrased would imply that if it was used in normal everyday conversation we would assume naturally that the bank teller answer means she is " just a bank teller " ....
isn't a large part of intelligence the able to figure out slightly distorted patterns ?
dont they test machine intelligence with similar things ?
so presuming that a person is asking you a question sincerely there is no mystery / fallacy IN THE ANSWER.
now to not realise this and try and explain the apparent mystery / error with quantum probability theory ....................................?
johnpopulati...
10/05/2009
Posts:1