LogicalMANIPULATIVE

Affirming the Consequent

What it is

A formal fallacy: from “if P then Q” and “Q,” concluding “P” — treating the occurrence of a predicted consequence as proof of the hypothesis that predicted it, when other hypotheses predict the same thing.

How it works

If the policy worked, unemployment would fall; unemployment fell; therefore the policy worked. The inference is invalid because the consequence may have many sufficient conditions, and the argument has ruled out none of them. It is nonetheless the everyday shape of confirmation: Popper (1959) built his account of science on the asymmetry that observing a predicted consequence cannot prove a theory (affirming the consequent) while observing a failed prediction can refute it (modus tollens). Wason (1968) showed how poorly people handle this: in his selection task most subjects turn the card that could confirm the rule and neglect the one that could falsify it, and Evans, Barston and Pollard (1983) found that believable conclusions make invalid forms more acceptable still. A confirmed prediction is weak support if it was unlikely on rival hypotheses, but the inference becomes a tactic when confirmation is presented as proof and the rivals are never stated. It drives the conspiracist's “they deny it, so there is a cover-up” and the marketer's “customers who use it are healthier.” The honest version is confirmation mistaken for demonstration. The knowing version chooses a consequence that any hypothesis would predict and calls it a test. The tell is a prediction fulfilled with no mention of what else would have fulfilled it.

Real-world examples

  • •Administrations of both parties claim that falling unemployment proves their policy worked; each opposition claims the same fall would have occurred anyway. Only a comparison with what rival hypotheses predict can adjudicate, and neither side usually offers one.
  • •Phishing exploits the inference: “If this were my bank, it would use my name and the last four digits of my account; it does, so it is my bank.” Attackers who have breached a database can fulfil the consequence.
  • •Diagnostic reasoning without base rates: “If you had Lyme disease you would be fatigued; you are fatigued; so you have Lyme disease” — the structure behind much self-diagnosis, since fatigue is predicted by nearly everything.
  • •The Kafka-trap form in conspiracy discourse: “If there were a cover-up, officials would deny it; they deny it; therefore there is a cover-up” — a consequence equally predicted by there being nothing to cover up.
  • •A start-up reports that customers who use its wellness app are healthier than non-users and concludes the app works; the healthier are also more likely to download wellness apps, a rival hypothesis that predicts the same observation.

Ethical guidelines

  • ●When you cite a fulfilled prediction as support, name the rival hypotheses that would also have predicted it, and say what distinguishes yours.
  • ●Prefer risky predictions: a test is informative in proportion to how likely it was to fail if you were wrong.
  • ●Do not present confirmation as proof; “consistent with” is the honest phrase, and “demonstrates” requires ruling alternatives out.
  • ●In product and policy claims, report the comparison group and what happened to it.

How to defend against it

  • ►Ask “what else would produce that?” — the one question the fallacy cannot survive. List the rival explanations before accepting the favored one.
  • ►Ask for the risky prediction: “What did your hypothesis predict that the alternatives did not, and did it happen?” A hypothesis that predicts only what everyone expected has not been tested.
  • ►Ask for the counterfactual or control: what happened where the cause was absent? If no comparison exists, the confirmation is anecdotal.
  • ►Flip the card (Wason): look for the observation that would falsify the claim, and ask whether anyone checked it.
  • ►For identity and security, treat fulfilled expectations (correct name, logo, account digits) as necessary but not sufficient; verify through a channel the sender does not control.

References

  1. Popper, K. R. (1959). The Logic of Scientific Discovery. Hutchinson
    The asymmetry between confirmation (affirming the consequent) and falsification (modus tollens) in scientific inference.
  2. Wason, P. C. (1968). Reasoning about a rule. Quarterly Journal of Experimental Psychology, 20(3), 273-281
    The selection task showing that people seek confirming cases and neglect falsifying ones.
  3. Evans, J. St. B. T., Barston, J. L., & Pollard, P. (1983). On the conflict between logic and belief in syllogistic reasoning. Memory & Cognition, 11(3), 295-306
    Belief bias: invalid arguments with believable conclusions are accepted more readily.
  4. Copi, I. M., Cohen, C., & McMahon, K. (2011). Introduction to Logic (14th ed.). Pearson
    The standard definition of affirming the consequent as an invalid form contrasted with modus ponens.
Last reviewed
Suggest a correction

Detect Affirming the Consequent in any text

Paste any message, email, or article into our free Manipulation Detector to see if Affirming the Consequent or other techniques are being used on you.

Try the Detector FreeCreate Free Account