Pattern-setting for improving risky decision-making

This study was originally conceived and designed in 2015 in close collaboration with Howard Rachlin. I wrote this manuscript and conducted new analyses for it after his death.

Failures of self-control are an all-too-familiar experience in human life, ranging from the mundane problem of eating too much of something tasty and getting a bellyache, to the potentially lethal crisis of severe drug addiction. But defining self-control, that is, defining behavior that constitutes a self-control success or a self-control failure, is fraught with difficulty. Rachlin (2017) proposed that self-control is behavior that is directed by higher-level, more temporally extensive goals, such as being healthy, in contrast to immediate goals, such as the short-term pleasure of another bite of food. This approach naturally fits into what Rachlin calls teleological behaviorism, the view that an organism's mind is best identified with the complex set of overlapping, often very long-term behavioral patterns that direct its behavior over its life (Rachlin, 1992; Rachlin, 2014a; see also Simon, 2023). Individual moment-to-moment actions, such as walking one step, occur as part of larger goal-directed behaviors, such as going to work, and patterns telescope outwards in this fashion until we reach the organism's highest-order, most abstract goals, such as being a good person or living a fulfilling life. To act with self-control, then, is to act in accordance with a long-term and important pattern rather than a competing shorter-term and less important pattern.

If an agent's long- and short-term behavioral patterns are in conflict, and obeying the longer-term patterns is to its benefit, then the agent may benefit from being committed to obeying the longer-term patterns. Thus the value of commitment devices, by which agents can reduce temptation or the opportunity to renege on a goal (Bryan, Karlan, & Nelson, 2010; Green & Rachlin, 1996; Rachlin & Green, 1972; Rogers, Milkman, & Volpp, 2014). For example, Giné, Karlan, and Zinman (2010) found that when smokers were given the opportunity to put money in a savings account that they only got back if they stopped smoking, they quit at a higher rate than controls. Although not forced to stop smoking, the smokers in the savings-account group had been given a monetary incentive to avoid this self-control failure. Bryan et al. (2010) discuss rotating savings and credit associations (ROSCAs), which provide an incentive to save money by requiring regular saving to get one's share of a communal pot. In a ROSCA, a number of savers meet periodically. At each meeting, everybody contributes an equal amount, and one person is awarded all the contributions. The awardee rotates from meeting to meeting, so eventually, every member gets an equal share.

Monetary incentives, however, aren't the only sort of commitment that ROSCAs provide. ROSCAs are small organizations, generally formed among people who already know each other, and that presumably lead to social pressure to continue saving. This softer kind of commitment, compared to losing money, may still be effective. Rachlin (2016) proposed an especially soft commitment strategy called pattern-setting (previously described in Rachlin, 2014b). The idea is that once a person has identified a self-control problem, such as an addiction to cigarettes, they begin by getting into the habit of recording how often they do the undesirable behavior. Then they set up a schedule of "pattern days" and "free days". On pattern days, they're obliged to do the behavior as many times as they did in the foregoing free day; for example, they must smoke the same number of cigarettes, no more and no less. Rachlin argues that by establishing large numbers of pattern days, the person will learn to associate each instance of the undesirable behavior on free days with its longer-term effect on pattern days. Thus, a smoker may opt to smoke less on a free day because one cigarette is no longer just one cigarette; instead, the one cigarette implies three or five or seven cigarettes smoked over the succeeding pattern days. If successful, this strategy would be a direct way to bring one's behavior into alignment with longer patterns, and thereby increase self-control, despite never requiring or even explicitly incentivizing reduction of the undesirable behavior.

Read, Loewenstein, and Kalyanaraman (1999) examined a situation similar to pattern-setting in a study of choice for movies. People were given a list of movies they could watch, some "highbrow" and some "lowbrow", on dates of their choice, one movie per day. They either selected each movie they would watch on the day they would watch it or selected three movies in advance to watch. Read et al. found that those choosing movies in advance were more likely to choose highbrow movies, supporting the idea that choosing in advance leads to more frequent "virtuous" (larger-goal-congruent) choices. Siegel and Rachlin (1995) examined various conditions in which pigeons chose between smaller-sooner (SS) and larger-later (LL) food reinforcers. When only one key peck was required per reinforcer, all subjects preferred SS. When 31 pecks were required, however, the pigeons came to prefer LL, even though only the 31st peck actually determined the reinforcer outcome, meaning that a pigeon who pecked the LL key 30 times in a row could still change back to SS on the 31st trial. Thus, behaving in larger units seems to have nudged the pigeons towards the larger reinforcer that required more patience.

The present study used a risky decision-making task based on Luhmann, Ishida, and Hajcak (2011) (to be replicated by Ciria et al., 2021) in which human subjects could choose between two gambles, one of which had a greater probability of paying out than the other but was only available after a delay. The analogy with the smoking-cessation pattern-setting scenario described earlier is that the better gamble is like not smoking: it may be a less appealing choice in the moment, but choosing it frequently has better long-term outcomes. In the current study, some subjects could choose freely on every trial, some were forced to repeat past choices according to one of two patterns, and some were forced to choose the better gamble on certain trials. It was predicted that over the course of the task, as they experienced the gamble outcomes and won the better gamble more often than the worse one, subjects would increasingly choose the better gamble in free-choice trials. Moreover, this effect was predicted to be stronger among subjects in the pattern conditions. It was predicted not to be stronger among those subjects who simply were forced to take the better gamble, because the theory behind pattern-setting (of individual free choices taking on greater weight) doesn't apply in this case.

Task code, raw data, and data-analysis code can be found at http://arfer.net/projects/pattern.

When people were obliged to repeat choices, they were more likely to choose the better option, that is, the one that required patience. This finding supports Rachlin's (2016) idea of pattern-setting: when their individual choices were given more weight by a pattern structure, the subjects seemed to prioritize a larger goal (earning money, whether real or hypothetical) over a smaller one (ending a momentary period of waiting). There was evidence for the effect of this pattern structure immediately, as well as a tendency for patterning to increase preference for the dominating option, Delayed, over the course of many trials, at least for the Within-Pattern condition.

Across two models, the estimated effects of the conditions can be summarized as follows. The Within-Pattern condition had a consistently positive effect on choices for Delayed, which increased over the course of the task. Across-Pattern had a weakly positive effect in Model 1 that mostly disappeared by the end of the task, contrasted with a strongly positive effect in Model 2. Force-Delayed had a negative effect in Model 1 that was weakened over the course of the task, contrasted with a positive one in Model 2. Control subjects chose Delayed less often over the course of the task. Meanwhile, payment had little relation with the outcome, and subjects who performed badly on the quiz were substantially less likely to choose Delayed.

Overall, the present results suggest that patterning can increase preference for a delayed dominating option. However, it isn't clear whether the shorter immediate pattern (Within-Pattern) is more or less effective than the longer but somewhat delayed pattern (Across-Pattern). Both patterns are still relatively short, leaving open the possibility that stronger effects might be obtainable with longer patterns, such as a single free choice constraining ten future choices. Furthermore, the findings for Force-Delayed are cryptic: perhaps being forced to choose Delayed made subjects want to try the other option more, or perhaps it helped them learn that Delayed was the right choice. The finding that Control subjects chose Delayed less often as the task progressed is contrary to hypothesis and suggests that it is difficult to learn which choice is best without guidance.

The fact that between-condition differences were visible to some extent immediately, as opposed to after a process of learning, is perhaps indicative of the self-knowledge and life experience that subjects brought to the task. People may under-correct for their tendency to make suboptimal decisions, but they are still savvy enough to take opportunities for correction in the first place, as by saving money that they will only be able to get back if they stick to a plan. Recall, for example, that Giné et al. (2010) found that smokers were willing to put money in a savings account that they would only get back if they stopped smoking. Similarly, people participate in ROSCAs although they require regular saving to get any payout. So, once subjects in the pattern conditions understood the special pattern constraints they would be subject to, they may have immediately seen the importance of not committing themselves to the worse gamble in pattern trials. Learning effects might have been strengthened if the pattern constraints had not been explained up front, and subjects had been obliged to learn on their own how the task worked. Such an arrangement would more closely resemble an animal study, but would also entangle what are in principle two distinct learning problems: discovering the task rules and adapting decision-making to those rules.

Considering that condition instructions seemed to suffice to produce the between-condition differences, it may be surprising how poorly subjects performed on the quiz. Only about half correctly answered all four comprehension questions about instructions they had just read. It was intended that the task program's own corrections and the reminder of trial-specific rules shown during each trial would minimize errors, but the especially low preference for Delayed among these poorly performing subjects suggests that they retained misunderstanding about the task. A limitation of the quiz is that it didn't cover condition-specific instructions, and it seems obvious that not all conditions are equally easy to understand. Stronger effects of the conditions, or a greater condition-independent preference for Delayed, might have been observed had subjects understood the task better.

Overall, this laboratory test of the viability of pattern-setting as a self-control technique succeeded, albeit not as fully as expected. Would pattern-setting then be effective for real-world self-control problems, like drug addiction? As with most laboratory studies, it is hard to predict how the effect sizes observed here would translate to substantive decision-making in the field, and whether they would be large enough to be practical. Only a direct empirical test could decide this, but it is worth considering how this study's methods differ from how Rachlin (2016) proposed pattern-setting might be used in the field. First, some subjects were offered hypothetical rather than real money, and studies investigating how hypothetical rewards differ from real ones in risky choice have produced inconsistent results (e.g., Barreda-Tarrazona, Jaramillo-Gutiérrez, Navarro-Martínez, & Sabater-Grande, 2011; Hinvest & Anderson, 2010; Horn & Freund, 2022, Xu et al., 2018; compare with the consistent finding that hypotheticality has little effect on intertemporal choice, as in Johnson & Bickel, 2002; Madden, Begotka, Raiff, & Kastern, 2003; Madden et al., 2004; and Lagorio & Madden, 2005). Second, subjects did not have to monitor and record their own behavior. The task took care of this for them. Not having to do the work of monitoring reduces the barrier to pattern-setting, and hence pattern-setting may be more difficult to use in practice than in this study; on the other hand, Rachlin (2016) suggested that the mere act of monitoring may help change one's own behavior. Third, and perhaps more important, subjects had no choice about obeying the pattern rules. It's one thing to smoke as many cigarettes as you want on Monday and say you'll smoke the same number on Tuesday; it's another to do on Tuesday what you said you'd do, which could itself become a self-control problem. A future test of pattern-setting might try telling subjects what the pattern rule implies they should do, rather than forcing them. Voluntary and consistent compliance with the pattern rule seems necessary for the potential of pattern-setting to be realized.

The initial instructions and the quiz were as follows:

Some quick notes before we begin:

• Please give the experiment your undivided attention. Doing something else (like checking your phone) during a waiting period would interfere with the purpose of the experiment.
• This experiment uses timers to make you wait for certain things. Don't use your browser's back button or refresh button on a page with a timer, or the timer may restart (in which case it will have the same length as before).

In this study, you'll complete a number of trials which will allow you to choose between two gambles, A or B. You can win real money from the gambles, which will be paid to you at the end of the study. You can't lose money from gambles. Right after you choose each gamble, I'll tell you whether or not you won the gamble.

[For unpaid subjects, the above instead read "In this study, you'll complete a number of trials which will allow you to choose between two gambles, A or B. You can win (imaginary) money from the gambles. You can't lose money from gambles. Right after you choose each gamble, I'll tell you whether or not you won the gamble. At the end of the study, I'll tell you your total winnings. Although no real money will be involved in this study, please try to make your decisions as if the gambles were for real money."]

Here's what the gambles look like:

[An example similar to Figure 1 of the main manuscript.]

The colored bars are just graphical representations of the chance of winning.

Notice that B has a higher chance of paying out. However, B isn't immediately available at the beginning of each trial. It will show as "[Not available yet]". You'll have to wait a random, unpredictable amount of time (sometimes short, sometimes long) for B to become available.

Choosing A will allow you to receive an outcome (either winning or not winning) without waiting, because A is available from the start of each trial. But choosing A won't let you complete the study any faster, because the time you would have waited for B, had you waited for it, will be added to the time you have to wait to get to the next trial (or to the end of the study). Any time you spend waiting before choosing A (although you don't need to wait before choosing A, as you do for B) will be credited towards reducing this wait.

Let's test your understanding.

1. Compared to B, A's chance of paying out is
   • lower [correct]
   • higher
   • the same
2. Which gamble gives you more money when you win the gamble?
   • A
   • B
   • They give the same amount of money [correct]
3. Which option can you choose as soon as a trial starts?
   • A [correct]
   • B
   • Either A or B
4. Which option will allow you to complete the study faster?
   • A
   • B
   • Neither; it makes no difference [correct]

Okay, one more thing before we begin.

You'll complete trials in blocks of 3.

[On a blue background:] In odd-numbered blocks (the 1st, 3rd, 5th, and so on), you'll see this background.

[On a yellow background:] In even-numbered blocks (the 2nd, 4th, 6th, and so on), you'll see this background.

The final line of the instructions varied by condition:

• Control: "The task works the same whether you're in an odd block or an even block."
• Within-Pattern: "Within each block, whether even or odd, you can choose either A or B on trial 1, but the task will force you to repeat that choice on trial 2 and trial 3."
• Across-Pattern: "In odd blocks, you can choose either A or B. In even blocks, the task will force you to repeat the series of choices you made in the previous block. So if in the 1st block you chose A, then B, then A, the task will force you in the 2nd block to choose A, then B, then A."
• Force-D: "In odd blocks, you can choose either A or B. In even blocks, the task will force you choose [sic] B on every trial."

Barreda-Tarrazona, I., Jaramillo-Gutiérrez, A., Navarro-Martínez, D., & Sabater-Grande, G. (2011). Risk attitude elicitation using a multi-lottery choice task: Real vs. hypothetical incentives. Spanish Journal of Finance and Accounting, 40(152), 613–628. doi:10.1080/02102412.2011.10779713

Bryan, G., Karlan, D., & Nelson, S. (2010). Commitment devices. Annual Review of Economics, 2(1), 671–698. doi:10.1146/annurev.economics.102308.124324

Ciria, L. F., Quintero, M. J., López, F. J., Luque, D., Cobos, P. L., & Morís, J. (2021). Intolerance of uncertainty and decisions about delayed, probabilistic rewards: A replication and extension of Luhmann, C. C., Ishida, K., & Hajcak, G. (2011). PLOS ONE. doi:10.1371/journal.pone.0256210

Cohen, J. (1994). The earth is round (p < .05). American Psychologist, 49(12), 997–1003. doi:10.1037/0003-066X.49.12.997

Cumming, G. (2014). The new statistics: Why and how. Psychological Science, 25(1), 7–29. doi:10.1177/0956797613504966

Gelman, A., & Stern, H. (2006). The difference between "significant" and "not significant" is not itself statistically significant. The American Statistician, 60(4), 328–331. doi:10.1198/000313006X152649

Giné, X., Karlan, D., & Zinman, J. (2010). Put your money where your butt is: A commitment contract for smoking cessation. American Economic Journal: Applied Economics, 2(4), 213–235. doi:10.1257/app.2.4.213

Green, L., & Rachlin, H. (1996). Commitment using punishment. Journal of the Experimental Analysis of Behavior, 65(3), 593–601. doi:10.1901/jeab.1996.65-593

Hinvest, N. S., & Anderson, I. M. (2010). The effects of real versus hypothetical reward on delay and probability discounting. Quarterly Journal of Experimental Psychology, 63(6), 1072–1084. doi:10.1080/17470210903276350

Horn, S., & Freund, A. M. (2022). Adult age differences in monetary decisions with real and hypothetical reward. Journal of Behavioral Decision Making, 35(2). doi:10.1002/bdm.2253

Johnson, M. W., & Bickel, W. K. (2002). Within-subject comparison of real and hypothetical money rewards in delay discounting. Journal of the Experimental Analysis of Behavior, 77(2), 129–146. doi:10.1901/jeab.2002.77-129

Kramer, A. D., Guillory, J. E., & Hancock, J. T. (2014). Experimental evidence of massive-scale emotional contagion through social networks. Proceedings of the National Academy of Sciences, 111(24), 8788–8790. doi:10.1073/pnas.1320040111

Kruschke, J. K. (2010). What to believe: Bayesian methods for data analysis. Trends in Cognitive Sciences, 14(7), 293–300. doi:10.1016/j.tics.2010.05.001

Lagorio, C. H., & Madden, G. J. (2005). Delay discounting of real and hypothetical rewards III: Steady-state assessments, forced-choice trials, and all real rewards. Behavioural Processes, 69(2), 173–187. doi:10.1016/j.beproc.2005.02.003

Luhmann, C. C., Ishida, K., & Hajcak, G. (2011). Intolerance of uncertainty and decisions about delayed, probabilistic rewards. Behavior Therapy, 42, 378–386. doi:10.1016/j.beth.2010.09.002. Retrieved from http://www.psychology.stonybrook.edu/cluhmann-/papers/luhmann-2011-bt.pdf

Madden, G. J., Begotka, A. M., Raiff, B. R., & Kastern, L. L. (2003). Delay discounting of real and hypothetical rewards. Experimental and Clinical Psychopharmacology, 11(2), 139–145. doi:10.1037/1064-1297.11.2.139

Madden, G. J., Raiff, B. R., Lagorio, C. H., Begotka, A. M., Mueller, A. M., Hehli, D. J., & Wegener, A. A. (2004). Delay discounting of potentially real and hypothetical rewards II: Between- and within-subject comparisons. Experimental and Clinical Psychopharmacology, 12(4), 251–261. doi:10.1037/1064-1297.12.4.251

Meehl, P. E. (1990). Why summaries of research on psychological theories are often uninterpretable. Psychological Reports, 66(1), 195–244. doi:10.2466/pr0.1990.66.1.195

Rachlin, H. (1992). Teleological behaviorism. American Psychologist, 47, 1371–1382. doi:10.1037/0003-066X.47.11.1371

Rachlin, H. (2014a). The escape of the mind. Oxford, England: Oxford University Press. ISBN 978-0199322350.

Rachlin, H. (2014b, July 24). If you're so smart, why aren't you happy? Retrieved from http://blog.oup.com/2014/07/psychology-self-monitor-abstract-particular

Rachlin, H. (2016). Self-control based on soft commitment. Behavior Analyst, 39(2), 259–268. doi:10.1007/s40614-016-0054-9

Rachlin, H. (2017). In defense of teleological behaviorism. Journal of Theoretical and Philosophical Psychology, 37(2), 65. doi:10.1037/teo0000060

Rachlin, H., & Green, L. (1972). Commitment, choice and self-control. Journal of the Experimental Analysis of Behavior, 17(1), 15–22. doi:10.1901/jeab.1972.17-15

Read, D., Loewenstein, G., & Kalyanaraman, S. (1999). Mixing virtue and vice: Combining the immediacy effect and the diversification heuristic. Journal of Behavioral Decision Making, 12(4), 257–273. doi:10.1002/(SICI)1099-0771(199912)12:4<257::AID-BDM327>3.0.CO;2-6

Rogers, T., Milkman, K. L., & Volpp, K. G. (2014). Commitment devices: Using initiatives to change behavior. JAMA, 311(20), 2065–2066. doi:10.1001/jama.2014.3485

Siegel, E., & Rachlin, H. (1995). Soft commitment: Self-control achieved by response persistence. Journal of the Experimental Analysis of Behavior, 64(2), 117–128. doi:10.1901/jeab.1995.64-117

Simon, C. (2023). "You can be a behaviorist and still talk about the mind – as long as you don't put it into a person's head": An interview with Howard Rachlin. Journal of the Experimental Analysis of Behavior, VOLUME(ISSUE), PAGES.

Standing, L., Sproule, R., & Khouzam, N. (1991). Empirical statistics: IV. Illustrating Meehl's sixth law of soft psychology: Everything correlates with everything. Psychological Reports, 69(1), 123–126. doi:10.2466/PR0.69.5.123-126

Wagenmakers, E. J., Wetzels, R., Borsboom, D., & van der Maas, H. L. (2011). Why psychologists must change the way they analyze their data: The case of psi: Comment on Bem (2011). Journal of Personality and Social Psychology, 100(3), 426–432. doi:10.1037/a0022790

Xu, S., Pan, Y., Qu, Z., Fang, Z., Yang, Z., Yang, F., … Rao, H. (2018). Differential effects of real versus hypothetical monetary reward magnitude on risk-taking behavior and brain activity. Scientific Reports, 8. doi:10.1038/s41598-018-21820-0