Cognitive bias

From Example Problems
Jump to: navigation, search
See also: List of cognitive biases.

Cognitive bias is any of a wide range of observer effects identified in cognitive science, including very basic statistical and memory errors that are common to all human beings and drastically skew the reliability of anecdotal and legal evidence. They also significantly affect the scientific method which is deliberately designed to minimize such bias from any one observer.

They were first identified by Amos Tversky and Daniel Kahneman as a foundation of behavioral economics.

Bias arises from various life, loyalty and local risk and attention concerns that are difficult to separate or codify. Tversky and Kahneman claim that they are at least partially the result of problem-solving using heuristics, including the availability heuristic and the representativeness heuristic.

Cognitive versus culture versus notation bias in physics

In physics, researchers are concerned with observer effects - which set clear limits on the process of observing as in Heisenberg's uncertainty principle. Quantum entanglement also introduces problems in some observational situations. These are all well-accepted foundations of 20th century philosophy of science and along with a few other such discoveries (like a universal maximum for the speed of light) form its core epistemology. Knowing these limits has helped develop a cognitive science by which humans might reasonably characterize the limits of their own perception.

However, bias does not end with cognition. How to interpret the data on what humans 'can' observe becomes controversial when there are few individuals capable of reproducing experiments and compiling new models. Culture arises.

The models and goals of experiments and the foundation ontology of which terms these are described often come into question as "normal science" nears a "paradigm shift", and considers changing this ontology, among other changes. Thomas Samuel Kuhn stated that the ontology or methodology accepted by prevailing scientists often provided a cultural bias - much like the one that previously guided human cognition of other species as being unfeeling or unconscious. Assigning all freedom of choice to the observer and none to the observed has been a common theme, and error, in all science prior to the 20th century.

Finally, there are notation problems. Aside from the neutrality of math as studied in the philosophy of mathematics, there are many shallow disputes:

For example, a quite different ontology problem identified by anti-reductionist physics is that there is just as much legitimacy describing the bonds or collections between particles as the first-class objects, as there is in the particle physics model composed of particles. Ontological ("what exists") and epistemological ("how you know") concerns often overlap, as they do in theology and philosophy where the terms originate - but descriptions of relations between cognitive bias, culture bias, and notation bias are as controversial as other questions of individual, social and instructional responsibilities - as most clearly visible in the wide range of economic studies of human capital.

Physics is only one science, but it shows the complexity of the problem, and the potential for cognitive bias to intervene even in discussions of culture or notation bias. Some theorists believe that a unified field theory is possible but only by unifying economics and physics into one field - a common dream of neoclassical economics but not widely shared by other subfields of economics.

Human observers versus chimpanzee observers

In his article "The Unreasonable Effectiveness of Mathematics in the Natural Sciences", 1960, Eugene Wigner noted in a footnote that it might be necessary to somehow determine whether other species saw the universe as we did, to get to objective answers free of human cognitive bias. He saw this as an epistemology question but others see it rather more literally:

Does science itself need primate testing? To date, it hasn't happened, except insofar as some chimpanzees have apparently learned to count: "Ai, 23, is no ordinary chimp. She has been learning linguistic and other skills at the Primate Research Centre since 1978. She is also a keen painter and has her own web site and e-mail address" just like other students. "She has learned how to use the Arabic numerals, 1 to 9, to represent numbers and her latest feat is to be able to memorise the order of five numbers."

In Nature, Tetsuro Matsuzawa claims that "Ai's performance shows that chimpanzees can remember the sequence of at least five numbers, the same as, or even more than, pre-school children," said the Japanese researcher who led the research, Other studies have shown that chimps have limited ability to add, but these remain unvalidated by other researchers. The capacity for math skills seems to suggest one avenue for reducing human cognitive bias.

Many other Great Apes have demonstrated mastery of small vocabularies - originally in American Sign Language but when these studies were themselves accused of observer bias, with pictographic keyboards. This is a clear but long-term route to dispelling some human cognitive bias - or "racial" bias:

Andrew Whiten, a primate expert from the University of St Andrews, said "In one sense, as a chimpanzee researcher, I'm not that surprised. Every year we find something else that chimpanzees can do which brings them one step closer to us." Regarding Ai's math skills: "We can remember a seven-digit number at first sight, our brain's 'magic number'. But here's a chimp remembering five - that's awfully close. And although their brain is large for a primate, it is only half the size of ours."

Conversations between humans and Great Apes seem sufficient to validate that they and humans share emotional and practical concerns, but not the sciences (physics, ecology) which we might expect to be most subject to human biases. Because these experiments do not take place in the wild, there are few attempts to establish a joint view of ecology (in one famous incident a chimp had to be scored down on an IQ test for choosing a tree over a house as shelter) or other subjects of joint interest. But the other sciences wait for a species that might be able to help us dispel biases arising from simply being human.

Weak, strong, and other anthropic principles

The most all-encompassing example of cognitive bias may be the anthropic principle: in its "weak" form, this speculation holds that we humans cannot observe any of the possible universes in which humans cannot exist, and therefore that the values of many fundamental constants of nature (e.g. the fine structure constant), and so of the nature of the Universe we see, are fixed by the fact that we exist to observe them. It is a tautology, and most scientists accept that peering into other universes is not possible - thus theories about them have serious problems of falsifiability.

The "strong form" of the anthropic principle (which most scientists do not accept) argues the reverse: that fundamental constants are constrained by cognition. How can this be, when mathematics predicts future events reliably?

One controversial argument, by Buckminster Fuller, held that cognitive bias arose not directly out of any fundamental physical constants, but rather out of the shared life, loyalty and locality imposed by living in one gravity well sustaining one biosphere - that rather than binding to the whole universe, intelligent creatures were bound emotionally and cognitively to their home planet. In his book Critical Path he concluded somewhat controversially that "gravity is love".

A range of less absolutist versions of this argument eventually led to Gaia theory in diverse forms, and the modern political movement of "Gaians" who deliberately assume the role of "Earth's Immune System" - and associated cognitive bias for taking such a limited role in such a very large universe.

Maths as shared abstract human cognitive bias?

However, arguing that planets or biospheres must hopelessly bias all aspects of our cognition strongly in their favor is not even the most extreme claim; some argue that whatever human cognitive bias exists is extended by culture into notations where it becomes institutionalized:

Many theorists have thought that mathematical notation itself must contain a strong human cognitive bias - that it cannot predict future events but rather future human cognition - a common theme in theology and in the philosophy of mathematics - and a few, led by George Lakoff, have attempted to create a cognitive science of mathematics to clearly bind together basic constants and operations in mathematics to human cognitive assumptions (about the body and its movements and negotiations). A more profound understanding of cognitive bias is expected to result from this.

However behavioral psychology and artificial intelligence studies in the 20th century had a similar agenda, and both more or less failed to explain more than a bare minimum of animal and planning behavior respectively.

See also: cognitive science, philosophy of mathematics, epistemology, list of cognitive biases.

External links

he:הטייה קוגניטיבית