Wason’s foundational research: the verification-falsification asymmetry

Wason’s (1968) selection task research established the foundational finding: people systematically seek information that confirms their existing hypotheses rather than information that would falsify them, even when falsification would be more informative about the hypothesis’s accuracy. The verification-falsification asymmetry is not a character flaw or a failure of intelligence; it is a systematic feature of how the human information-seeking system works. Confirming evidence feels like learning; disconfirming evidence requires deliberate effort to seek and accept.

Nickerson’s (1998) comprehensive review of the confirmation bias literature confirmed that the asymmetry operates across all domains and all levels of expertise. Domain expertise does not reduce confirmation bias; in some research it intensifies it, because expertise deepens the commitment to the expert’s existing framework — which is the framework the confirmation bias then seeks to confirm. The entrepreneur who has spent three years developing a specific market hypothesis has three years of accumulated confirmation-seeking invested in that hypothesis; the confirmation bias protecting it is correspondingly strong.

The practical implication is that thorough research conducted by the entrepreneur themselves about their own business is, by default, confirmation-biased research. Not because the entrepreneur is dishonest but because the information-seeking behaviour that confirmation bias produces — asking questions that elicit validation, selecting data sources that tend to confirm the hypothesis, weighting confirming evidence more heavily than disconfirming evidence — is experienced from the inside as thorough, objective research rather than as the selective attention it is.

The Kunda motivation mechanism: why high stakes intensify the bias

Kunda’s (1990) motivated reasoning research established the activation mechanism that makes confirmation bias most damaging precisely when it is most costly. Motivated reasoning is the tendency to reason toward conclusions that serve the reasoner’s goals or protect their beliefs — and its intensity is proportional to the personal investment in the conclusion being protected.

The entrepreneur who has committed their financial resources, professional reputation, and identity to a specific business hypothesis has maximum personal investment in that hypothesis being accurate. The confirmation bias that motivated reasoning produces is therefore most intense in the evaluation of that hypothesis — which is the evaluation where accurate, unfiltered information would be most commercially valuable. The investment that makes accurate information most important is the same investment that most powerfully activates the bias toward finding confirming information instead.

This explains the specific pattern that Baron’s (1998) entrepreneurial cognition research documented: entrepreneurs show significantly stronger confirmation bias than matched control samples in opportunity evaluation. The explanation is not that entrepreneurs are more biased than other people in general; it is that the opportunity evaluation context maximises the personal investment that Kunda’s mechanism predicts will maximise the confirmation bias intensity.

Three specific manifestations in entrepreneurial practice

The customer interview that asks “do you like this?” is the most commercially costly single expression of the confirmation bias in entrepreneurship. The question is designed to elicit validation — and it typically succeeds, because the customer’s social dynamics favour agreement over honest evaluation. The Fitzpatrick Mom Test methodology was designed specifically to address this pattern: replacing validation-seeking questions with questions about past behaviour (“tell me about the last time you tried to solve this problem”) that elicit actual experience rather than socially compliant opinions about a hypothetical product.

The market analysis that reports the most optimistic projection from a range of analyses is the same mechanism applied to data rather than interviews. The entrepreneur who selects the TAM figure from the source that produces the largest number, the growth rate from the analysis with the steepest trajectory, and the competitive landscape analysis that most minimises the established competitors’ genuine strengths is not fabricating the research; they are applying the confirmation bias’s selective attention to real data sources and producing a real but systematically biased picture.

The competitive analysis that identifies only the weaknesses of established competitors — missing their genuine strengths, the customer loyalty their product has earned, the distribution advantages their scale provides — is the strategic confirmation of the business’s differentiation hypothesis through selective attention to the evidence that confirms it.

Why the bias is specifically resistant to self-detection

Pronin, Lin and Ross’s (2002) bias blind spot research established the specific feature of confirmation bias that makes it most difficult to address: the bias is specifically resistant to self-detection because the information-seeking behaviour it produces feels like thorough research rather than biased research. The entrepreneur who has conducted extensive customer interviews, reviewed comprehensive market analyses, and built detailed competitive landscapes has subjective evidence of thoroughness that makes the confirmation-bias-driven selectivity invisible. The research felt thorough; it looked thorough; it was thorough — and systematically biased in ways that thoroughness alone does not prevent.

This is why the structural interventions that the research supports are more effective than the metacognitive ones. Deciding to seek disconfirming evidence, while genuinely intended, runs against the verification-falsification asymmetry that is generating the bias. Structural interventions that force the disconfirming search — regardless of the researcher’s natural information-seeking inclinations — are more reliably effective.

The structural interventions the research supports

Klein’s (2007) pre-mortem technique is the most directly evidence-based structural confirmation bias reduction tool: explicitly imagining that the venture has failed, and working backward to identify the most plausible causes, forces the search for disconfirming evidence that the confirmation bias naturally suppresses. The pre-mortem does not require overcoming the bias through willpower; it redirects the information-seeking behaviour toward a different question — what went wrong? — that the confirmation bias cannot easily answer with confirming evidence.

The Tetlock superforecasting practice of explicit, scored probability tracking significantly reduces confirmation bias over time because the scoring system makes the bias’s consequences visible and quantified. The forecaster who consistently assigns 90% probability to hypotheses that turn out to be true 60% of the time can see the systematic overconfidence in the record — which the confirmation bias cannot easily rationalise away from.

Books worth reading on this

You Are Not So Smart by David McRaney is the most accessible available popular account of confirmation bias and related self-deception mechanisms — covering how the confirmation bias operates across everyday thinking, the specific social and cognitive conditions that intensify it, and the ways that awareness of the bias changes the relationship with it without eliminating it. McRaney’s account of why smart people show the bias at least as strongly as less sophisticated thinkers — and often more so, because intelligence produces more sophisticated rationalisations of confirmation-biased conclusions — maps directly onto the Kunda motivated reasoning and Nickerson expertise-confirmation findings this article describes.

If the dynamics described here are significantly affecting your wellbeing, speaking with a psychologist is the right next step. UK: Samaritans (116 123, free, 24/7). Mind (0300 123 3393). BACP: bacp.co.uk/search/Therapists. Crisis Text Line — text HOME to 741741 (US, UK, Canada, Ireland). International: internationaltherapistdirectory.com.

This article is for educational and informational purposes only. Sources: Wason, P.C. (1968), Reasoning about a Rule, Quarterly Journal of Experimental Psychology, 20(3), 273–281. Nickerson, R.S. (1998), Confirmation Bias: A Ubiquitous Phenomenon in Many Guises, Review of General Psychology, 2(2), 175–220. Kunda, Z. (1990), The Case for Motivated Reasoning, Psychological Bulletin, 108(3), 480–498. Baron, R.A. (1998), Cognitive Mechanisms in Entrepreneurship, Journal of Business Venturing, 13(4), 275–294. Fitzpatrick, R. (2013), The Mom Test, Robfitz. Pronin, E., Lin, D.Y. & Ross, L. (2002), The Bias Blind Spot, Personality and Social Psychology Bulletin, 28(3), 369–381. Klein, G. (2007), Performing a Project Premortem, Harvard Business Review, 85(9), 18–19. McRaney, D. (2011), You Are Not So Smart, Gotham Books. Syed, M. (2015), Black Box Thinking, John Murray.