Conjunction fallacy

Cogito ergo sum Logic and rhetoric

Key articles

Logical fallacy Syllogism Argument

General logic

Modus tollens Presentism Type theory Tone argument Lógica Generalização precipitada

Bad logic

Gish Gallop Ipse dixit Red-baiting Counterfactual fallacy Appeal to nature Media was wrong before v - t - e

The conjunction fallacy is a formal logical fallacy and a cognitive bias, among other things associated with the sway of stereotypes over thought, and belief in the paranormal and in conspiracy theories. Depending on the context, it may also be referred to as the conjunction effect.

For two events A and B having independent probabilities, the probability that both occur together is never larger than the probability that one of them (regardless of the other) does. This can be written as, $\Pr(A\land B)\leq \Pr(A)$ and $\Pr(A\land B)\leq \Pr(B)$. Anything that contradicts this mathematical relationship is an example of the fallacy.

Forms

It can have several forms. The most well-known is the below.

P1: A seems unlikely.

P2: B seems likely.

C: A and B seem more likely than A.

The most well-known example of this is the "Linda problem" by Amos Tversky and Daniel Kahneman, a question which they had people answer in their research – the majority choosing the second option.

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?

1. Linda is a bank teller.
2. Linda is a bank teller and is active in the feminist movement.

Variations in wording may affect how susceptible people are to this cognitive bias. Tversky and Kahneman found that changing the first alternative to "Linda is a bank teller whether or not she is active in the feminist movement" resulted in fewer making the error, though a majority of 57% still did. Studies have also varied the presentation in various ways to guard against other explanations for how people rank the options than the proposed cognitive bias.

Another fairly general form that ties the role of heuristics or intuitive judgement into it is the below. It captures more items from Tversky and Kahneman. For example, in 1980, respondents judged the scenario of the US breaking off diplomatic relations with the Soviet Union after the latter invades Poland more probable than merely a breaking off of US-Soviet diplomatic relations with nothing more specified. Tversky and Kahneman explain such skewed judgment in terms of a representativeness heuristic guiding the mind toward irrational conclusions.

P1: A seems so-and-so likely and gives some subjective impression.

P2: B seems so-and-so likely and gives some subjective impression.

C: A and B together give a balanced or more realistic impression and so seem most likely.

The point here being that the subjective "realism" comes from the representativeness heuristic – which is swayed by stereotypes and by representative ideas and images which have formed through taking in all kinds of impressions, irrespective of rational evaluation.

Without a choice of conjunction

A conjunction effect can be seen even when a person only considers one option, rather than a choice involving having or not having the logical conjunction. It's then typically not called a fallacy but rather the "conjunction effect". In other words, there's not always any formal argument structure involved in errors of this kind, rather the intuitive impression when considering some thing is just biased. Tversky and Kahneman actually studied this first.

For example, in a variation on the "Linda problem" study, each individual test subject is given one of two lists of options, each list lacking either the main option with a conjunction or the main option without it. When ranking how probable things seem, each person only considers one of those two options of interest, and yet, the effect is shown in the relative rankings of how probable the two options are judged along with more options across the whole study.

Class-inclusion problems

A related form of faulty reasoning, shown to be common in the thought processes of small children, is the group of class-inclusion problem, going back to the child psychologist Jean Piaget. If, for instance, children are shown a picture of 7 cows and 3 horses and asked whether there are more cows or animals, they may reply, "more cows", although they are aware of the fact that both horses and cows are animals (this example of the error was shown to occur until around age 10). This error shows a failure to apply knowledge of hierarchies or to reason in a logical fashion about hierarchies, instead only considering one level (cow) without considering the others. It is also connected with riddles based on semantic ambiguity, such as "What barks but is not a dog?" "A bitch" (where "dog" refers ambiguously to the male of the animal or all canines); children are able to appreciate such riddles from around age 8, although even adults may not always get the answer.

In assorted woo

There's research suggesting that people who believe in paranormal ideas are particularly susceptible to the conjunction fallacy, and that the same also goes for conspiracy theorists. Intuitively, the conjunction fallacy would seem not only to make it easier to accept bogus ideas, but also to possibly promote crank magnetism by making it easier to accept more elaborate combinations of all kinds of crank ideas. We can only speculate on such matters here at RationalWiki, though.

To speculate on how the conjunction fallacy may work in relation to woo teachings, it may be useful to consider which stereotypes and cliché ideas and images are popular – more generally what the representativeness heuristic may latch on to. For example, to a person who thinks the X-Files are close to reality and really wants to believe, what kinds of things may seem representative of reality? Or, to someone in a cult which teaches that suppressed knowledge is all-important for salvation? What kinds of additions to "this and …" may seem to boost how likely it all is to such people? There could possible exist interesting examples to be found in the world of conspirituality.

From looking at examples of crank magnetism in action, a tongue-in-cheek suggestion for an aspiring cult leader could be, in order to suck in those predisposed to be believers in wild combinations of claims added to claims: claim a lot, at once and over time. If you assert some things to your would-be believers without evidence, and they don't care much for that message, then try piling on more, not less. Conjure up emotional images, draw upon stereotypes and cognitively accessible clichés and fictional tropes, the kind of stuff people tend to intuitively view as fitting together as part of what they want to be real, and those predisposed to this fallacy may be able to swallow a lot of claims in conjunction all in one go. Of course, it can't necessarily all be chalked up to this fallacy when people do swallow such teachings, though – the psychology of belief is complicated.

See also

• Cognitive bias
• Crank magnetism
• Logical fallacy