Modification of slot-machine preferences through the use of a conditional discrimination paradigm.

Adults played a computer slot machine that offered 'let-it-ride' bets. The researchers first watched how often each person risked the bet. Then they taught a short rule: either 'keep betting, the machine is due' (a faulty rule) or 'walk away, the odds are fixed' (a correct rule). They used a conditional-discrimination procedure so each rule was tied to colored lights on the screen.

Faulty rules made risky bets shoot up. Correct rules pulled them back down, but most players still bet more than they did at the start. Rules alone moved the needle, yet they did not fully undo the earlier risky pattern.

Belisle et al. (2019) extend these results. They added a five-minute 'defusion' exercise right after the same kind of conditional-discrimination training. With that small add-on, only 4 of 10 adults later showed a gambling bias, a cleaner win than the mixed picture seen here.

Schlinger (1990) laid the groundwork. That paper told us to call something a 'rule' only if it actually changes the contingency, not just describes it. Giallo et al. (2006) tested the idea by showing that a single sentence could swing real money choices.

Stüttgen et al. (2024) zoom in closer. They tracked trial-by-trial shifts in perceptual choice and found reinforcement history steers each click. The slot study did the same—each let-it-ride decision was a trial—so both papers show moment-to-moment control, one with words, one with numbers.

If you teach a client a safety rule—like 'stop when the timer beeps'—remember that later misinformation can partly overwrite it. Build in quick defusion or practice rounds, as Belisle et al. did, to lock the correct rule in place. Check early and often; words help, but they are not bulletproof.