January 4, 2014

Informed Intuition > Pure Logic, Reason + No Information = Fallacy?

This content has been cross-posted to Tumbld Thoughts.

The peer-review committee for pure rationality. COURTESY: [1]

My notes on logical fallacies: perhaps they are not as bad as you think. People can make what are clearly errors in logic, and sometimes such fallacies are used as decision-making heuristics or cultural blends. This helps us make difficult decisions in the absence of information, or make sense of situations with little precedent.

This is like the 12-step program for skeptics and humanists (or those who aspire to these values). Much like 12-step programs, they leave a lot to be desired. These rules are largely naive of propagandist techniques, and the innate cognitive and cultural biases of their readers. The rhetoric argument does not fare with respect to this list. Fallacies on this list that I have an issue with:

1) Special pleading (and appeal to emotion): in cases where people fail to understand the context of a decision, special pleading might help to offset the damage done by a purely logical decision. Legal decisions that do not take in special cases (e.g. Grandfather clauses) are particularly of note.

2) Black-or-white: if decision-making were entirely deliberative (e.g. purely logical), we would never arrive at a decision. In this sense, decision-making must include a impulsive (or emotional) component. 

3) Ad hominem: while attacking the person rather than the argument is a convenient way to win an argument, this idea also assumes that people always argue in good faith and from a position of pure objectivity. This leaves no room for a theory of motivation, particularly when an argument has a thinly-veiled ulterior motive.

4) Slippery Slope: in cases of ambiguous moral or logical clarity, the slippery slope might actually help us clarify boundaries between one state and another. Without this boundary, human cognition is left without a reference point, which does not allow for clear (and culturally-relevant) decisions to be made.

5) Ambiguity: ambiguity is a necessary condition of a living argument. In cases where ambiguity is resolved, argument or belief/rule system becomes constricted. Allegorical arguments depend on ambiguity to remain relevant -- perhaps this is simply support for the ambiguity fallacy, but allegories are important devices in abducing (e.g. logical abduction) new logical relationships.

6) Strawman: "misrepresentation" of an argument is often in the eye of the beholder. People tend to extract heuristics in dealing with complex arguments, so it is hard to not construct a strawman (unless it is exceedingly flimsy, as with most intelligent design endeavors). 

Unless an argument is painstakingly recapitulated, any "elevator talk" length summary is bound to fail. And sometimes arguments are inherent to one's belief system -- in fact any criticism in this case could be viewed as a misrepresentation. In any case, strawman-type approaches can be used to set up improvements to an argument.

Not so fast, deduction fans......

Now here are some new fallacies that I have come up with. These are based on personal experience, both in human interactions and artificial intelligence. They are a bit more nuanced and specific than the fallacies presented in the "12-step program" model, but then again it speaks to some of my critiques above.

1) The "economic argument"/argument from efficiency: resource allocations that benefit me or my social group are superior, and can be extended to efficiency criteria. 

2) The correlative argument: things that co-occur are always significant. The problem is not discussed in the framework of complex, multivariate causality.

3) Argument from exemplar, normalization fallacy: similar to the correlative argument, but runs in the other direction. In this case, the argument is made from a single example.

In some cases, while the argument is made from extended observation, those observations do not map to the natural phenomenon well. Alternately, comparing phenomena that do not have the same underlying statistical distribution is an example of the normalization fallacy.

4) False consensus: consensus (meeting of the minds, political coalitions, peer-review) always puts you in a better place than where you started. A variant of the normalization fallacy, but involves the assumption that intellectual triangulation will solve any problem.

5) Argument from extreme relativism: when every culture is correct, no matter how morally repulsive the practice. This comes from a misunderstanding of cultural relativism: relativism is not about values, but about the intersubjectivity of cultural variants. In other words, these variants cannot be understood in isolation, only in the context of other practices.

6) Argument from moral superiority: an argument that is rooted in moral superiority (using partially or questionably factual information to intimidate). The goal of such an argument is to reform, prosletyze, or otherwise morally manipulate the intended target.

7) Highly-contingent statistic fallacy: statistics that are the most extreme in recorded history, or the first time a double play was turned in the 5th inning by a left-handed second baseman at night. The superlative is misleading, because the situation is either highly-artificial or not conducive to replication.

Different Ways of Explaining

To conclude, I will demonstrate how there are different ways of explaining. Is one specific type always superior, or is it context-dependent? Or do the abductive and deductive approaches have their own unique advantages?

Ghosts in the machine....

Who's better at explaining things, a novice with an interest and a creative mind, or an expert at really complex concepts [2]? Here, we have an example of the former (Bjork explaining how a TV works in three minutes) and the latter (Hiroshi Ishiguro and other robotics experts explaining the uncanny valley in one minute). And, of course, Spock can address both sides of the equation in one quick caption.

Robots and Humans. Theory of Mind but no context-dependence. Weird and Unnerving.

[1] This is a list of 24 common logical fallacies, courtesy of The Skeptics Guide to the Universe and Yourlogicalfallacyis.com (Jesse Richardson, Andy Smith, and Som Meadon). Also, most of these are individually found on Wikipedia with a more detailed explanation.

[2] For an interesting example of how readers of a magazine for statistics professionals explained the Monty Hall problem to a general audience: Reader's Challenge: the Monty Hall problem. Significance magazine, October, 32-33 (2013).

No comments:

Post a Comment