Disjunction reasoning
Inference about 'or' — which people find harder and often under-explore.
What it means
Disjunction reasoning concerns inferences over alternatives joined by 'or,' where correct conclusions require considering each possibility, yet people tend to under-represent the full set of cases. The disjunction effect, a violation of Savage's sure-thing principle, shows people declining a choice they would make under either resolution of an uncertainty simply because the uncertainty is unresolved. Mental model theory attributes the difficulty to a 'principle of truth': reasoners represent only what is true in each possibility and neglect the implicit false cases, generating illusory inferences with disjunctions. Disjunctive thinking is effortful because holding multiple models in mind taxes working memory. Its failures appear in hypothesis testing, contingency planning, and decisions made under unresolved uncertainty.
Two distinct failures under one word
The term bundles two problems that share the word 'or' but come from different places. One is decision-theoretic: the disjunction effect, where a person who would act under either resolution of an uncertainty still refuses to act while it is unresolved. The other is logical: illusory inferences, where reasoning over disjunctive premises produces conclusions that feel valid but are not. The first is a failure of consequential choice, the second a failure of deduction. They are worth separating because they have different cures. The decision effect softens when you make the person spell out what they would do in each case; the logical illusion softens only when the reasoner is forced to represent the possibilities in which the premise is false, not merely those in which it is true.
What the evidence shows
The disjunction effect is real but fragile. Tversky and Shafir's original two-stage gamble, where people decline a second bet while the first is unresolved, has resisted replication: Kuhberger, Komunska and Perner found no significant effect across several payoff schemes, including real money. A large 2021 replication by Ziano and colleagues (N=890) recovered the effect for the 'paying to know' vacation problem but found the choice-under-risk version much weaker than reported. So the phenomenon is paradigm-dependent, strongest where an uninformative payment buys premature certainty and weakest in repeated gambles. The illusory-inference work is more robust: Johnson-Laird and Savary showed people endorse invalid conclusions from disjunctions at high rates, and Khemlani and Johnson-Laird replicated and extended this with embedded disjunctions.
Why 'or' taxes the mind
A disjunction names more than one way the world could be, and holding several possibilities open at once is expensive. Mental model theory explains the errors with a 'principle of truth': to spare working memory, people represent only what is true in each possibility and leave the implicit false cases unrepresented. That shortcut is usually harmless but occasionally produces a conclusion that the neglected cases would have blocked. A second complication is that ordinary language 'or' is ambiguous between inclusive and exclusive readings, so people must also guess whether both alternatives can hold together. The combination, more models to track and an unstated scope, is why disjunctive premises reliably slow people down, raise error rates, and invite premature closure on the first branch examined.
Where it shows up and how to blunt it
Disjunctive under-exploration surfaces wherever alternatives must be carried forward: differential diagnosis, fault-finding, contingency planning, and legal or investigative reasoning where a fact could have several sources. The common pattern is that a second possibility is named aloud and then quietly dropped once the first is checked. The practical countermeasure follows from the mechanism. Force the omitted models to become explicit: write out each branch as a separate line, including the case where both alternatives hold, and state what you would decide under each before information arrives. In deductive settings, prompting reasoners to consider the falsifying possibilities, rather than only confirming ones, sharply reduces illusory inferences. The cost is real cognitive effort, which is why the correction has to be built into procedure rather than left to good intentions.
Examples
Students say they would take a vacation if they passed an exam, and also if they failed it, yet defer booking until the result is known — even though the outcome should not matter.
Asked whether the launch survives if the supplier is late or the budget is cut, a team reasons through each case alone and never builds the third one where both happen.
Told the failure is either the cache or the network, an engineer checks the cache, finds nothing, and stops. The second possibility was named but never actually held in mind.
An investor would hold the position whether the merger closes or collapses, yet sits in cash until the shareholder vote, paying in lost time to postpone a decision the outcome does not change.
A recruiter, told a resume gap is either caregiving or a failed venture, screens the candidate out after picturing only the venture, with the caregiving branch listed but never actually weighed.
First described in Tversky & Shafir (1992); Johnson-Laird & Savary (1999).
Key references
- Ziano, I., Kong, M. F., Kim, H. J., Liu, C. Y., Wong, S. C., Cheng, B. L., & Feldman, G. (2021). Replication: Revisiting Tversky and Shafir's (1992) Disjunction Effect with an extension comparing between and within subject designs. Journal of Economic Psychology, 83, 102350. doi.org/10.1016/j.joep.2020.102350
- Khemlani, S., & Johnson-Laird, P. N. (2009). Disjunctive illusory inferences and how to eliminate them. Memory & Cognition, 37(5), 615-623. doi.org/10.3758/MC.37.5.615
- Kuhberger, A., Komunska, D., & Perner, J. (2001). The disjunction effect: Does it exist for two-step gambles? Organizational Behavior and Human Decision Processes, 85(2), 250-264. doi.org/10.1006/obhd.2000.2942
- Johnson-Laird, P. N., & Savary, F. (1999). Illusory inferences: A novel class of erroneous deductions. Cognition, 71(3), 191-229. doi.org/10.1016/S0010-0277(99)00015-3
- Tversky, A., & Shafir, E. (1992). The disjunction effect in choice under uncertainty. Psychological Science, 3(5), 305-309. doi.org/10.1111/j.1467-9280.1992.tb00678.x