site banner

Friday Fun Thread for September 16, 2022

Be advised; this thread is not for serious in depth discussion of weighty topics, this thread is not for anything Culture War related. This thread is for Fun. You got jokes? Share 'em. You got silly questions? Ask 'em.

6
Jump in the discussion.

No email address required.

Are non-contingent rewards (NCRs) an effective way of incentivizing behavior? I was surprised to read that there was controversy on this, then surprised to learn I had wrong beliefs about it.

A reward can be either contingent or non-contingent. Contingent means the reward can only be obtained from a behavior (or most easily obtained, more likely to be obtained). A non-contingent reward is a reward that comes after a behavior, but can also be obtained (easily) without the behavior. For an example, the contingent rewards of getting a good grade are pride, social esteem, competitive victory, future opportunities, and the intrinsic feeling of mastery involving the subject and the goal. A non-contingent reward might be treating yourself to a steak dinner afterward, buying nice clothes, certain kinds of pride that do not contingently follow from the behavior (general self-praise). You can enjoy NCRs without the preceding behavior, whereas CRs require the behavior.

Humans appear to be the only animal that attempts non-contingent rewards. Studies in pigeons that attempt to introduce NCRs by creating a habit involving reinforcement/punishment and then removing the reinforcement/punishment find that as soon as pigeons realize they can eat without the behavior, they do so rapidly. From the human studies I’ve read, NCRs are at best insignificantly beneficial, and this benefit is probably from the invisible residue of CRs.

A lot of things that people do to “reward themselves” may appear beneficial because they remind the person of the contingent rewards that follow, and not actually because the NCR is effecting anything. For instance, if you praise yourself with good self talk after a desired behavior, this is an NCR; however, the very words we use in this self-praise are also reminders of the whole world of social rewards that may come from the behavior. Let’s say you praise yourself after running a mile. If you tell yourself “great job, you did so well”, that is construed in the literature as an NCR. However, these words do not exist in isolation, and they will remind the runner of past social praise and the possibility of future social praise, which are contingent rewards. Self-talk is the most effective NCR in the literature and is also inextricably jumbled up with CRs; the NCR theory does not assert that cues to CRs increase behavior (of course they do), and so some self-talk should not be considered an NCR.

NCRs may also appear effective because they reduce baseline pain or increase baseline pleasure. If I have a yummy smoothie every time I work out, this would be an NCR. But the yummy smoothie might not be increasing my desire to run, and instead be allowing me to act on my desire to run by sufficiently reducing baseline pain. IE we’re in a good mood and not hungry we might be more likely to best somewhat unpleasant things.

tl;dr it is probably impossible to reward yourself; best to cling to the rewards that require your desired behavior

Most studies I've seen have such artificial settings, it's hard to take their conclusions at face value (if at all). Do you have any persuasive studies in mind? Or even better, just data sets/ narrative reviews of human behavior in the wild, w/t models and theories?

https://www.sciencedirect.com/science/article/pii/S0005789417301004

https://www.pcmasolutions.com/PDF/YouCantReinforceYourselfweb.pdf

These were persuasive enough for me. And just some thoughts influenced my opinion: what is the likelihood that humans are the only animal that can trick themselves with a non-contingent reward? If I can trick myself with a non-contingent reward, I can just as well trick myself into no longer desiring the target behavior and instead just enjoying the NCR. Or, if I can trick myself into believing the NCR is the result of my behavior, I may as well trick myself into believing any enjoyment in life is a result of my behavior, so the NCR is unnecessary. The very idea of an NCR, when you think about it, requires a burden of proof that is not satisfied by any study I could find, because it doesn’t make sense given what we know from animal psychology.

There’s also that, if they worked, we would see their usage among top performers in a variety of domains, given how easy they are to implement over contingent rewards. We would expect all the best students to have the most NCRs at the end of a successful study session, but we don’t really find this. If anything, introducing NCRs (especially social media, video games) within a study session is highly discouraged, because top students spend hours in deliberate study sessions which are usually not NCR’ed. Studies on top violin performers show that they are more likely to nap after an afternoon session, whereas NCR theory holds that they would be better of eating skittles or watching their favorite show. The top students I know would care obsessively about their grades without trying to self-incentivize, and they would do things that enhance the salience of grades like making detailed plans and hoping in relevant future opportunities.

what is the likelihood that humans are the only animal that can trick themselves with a non-contingent reward?

Humans are also the only (or one of the very few) animals that can solve the Buridan's Ass problem. Lev Vygotsky in collaboration with Pavlov ran a series of experiments: a hungry dog was put in a room with food, separated from it by a section of metal floor giving it mild electric shocks. Dogs behaved according to Behaviorist theory: since at every moment the hunger stimulus is weaker than the stimulus from the electrified floor, so the dog cannot cross it, despite the fact that the unpleasantness of hunger over long time massively outweighs the unpleasantness of crossing the floor. IIRC dogs responded by either getting into a catatonic state or getting enraged, which sounds like second best possible responses to such kind of situations.

An adult human of course solves the problem with remarkable ease: you just decide that you want to cross and this internal stimulus adds its weight and lets you cross the electrified section. Vygotsky also claimed that we can see development of this ability in children or primitive peoples, where it first requires an external stimulus like divination (flipping a coin), but then gets internalized.

NCRs seem to fall into roughly the same category of external willpower amplifiers. So not surprising that it's not only uniquely human but also that highly successful people don't need it any more.

That’s an interesting study. There are other possibilities to human uniqueness, though: humans in a lab know that they are actually safe, and they know that the shock is removed when the food is accessed. An animal lacks the human knowledge of that the pain is transient and surpassable. So I wonder how transferable the experiment is from animal to human. Perhaps we would have to have an experiment where an animal sees another animal accessing the food and being okay. Otherwise the animal has no way of knowing that the pain is harmless and transient.

There are lots of mental disorders in humans where the pain surpasses the reward, like in anxiety disorders, OCD, and eating disorders; in these cases the subject has no actual belief in the transience of the pain, and so could literally starve to death from anorexia or hikkomori-ism before attempting a reward.

Some dogs definitely figure this out w.r.t. those wireless electric fence/shock collar thingies; basically in addition to shocking them as they approach the boundary wire, you need to also train them that the shock (+ audible cue) means they need to turn around and go back. Otherwise they just push past the boundary and enjoy their freedom.

Yeah, it's unfortunate that the whole thing kinda got forgotten due to the Iron Curtain, so I haven't heard about any attempts to replicate and further investigate it these days.

This stuff meshes very well with a lot of stuff when you begin to look at things from that angle. Consider for example this cute video of children subjected to the marshmallow test: https://youtube.com/watch?v=QX_oy9614HQ&t=16 . Forget about the controversy about whether it actually correlates with important life outcomes all that well, the interesting thing is that we can see how the children want to avoid eating the marshmallow, how they employ various external (to the mind) willpower aids like covering their mouth with their hands etc, and how we know that as adults we wouldn't need to.

Thank you!

Unexpectedly, very few studies met our inclusion criteria, despite the fact that we included both “self-incentive” and “self-reward” as search terms.

They took very narrow inclusion criteria. It might genuinely reflect their narrow research interest, but there is a vast overlapping terminological mess research on behavioral change and goal pursuit, which uses other notions: intrinsic/extrinsic motivation, self-control / effortless control / impulse control, reward processing/ neural economics, etc. After brief unsuccessful attempts to "get the whole picture", I now read mostly about specific techniques, which are easier to test empirically, and that seem to work for me. Are you researching out of scholarly interest or to enhance personal performance?

In equal parts interest and personal gain. Any recommended readings?

The following is a personal sample, I am not an expert. I prefer bottom-up control: to craft your environment and schedule in advance, so as to get the right stimulation and avoid dangerous cues - with minimal effort. This paper with sophisticated title is about addictive behaviors in general, I find it conceptually useful. In good environment attention still fluctuates and has to be restored: this paper outlines neurobiological mechanisms of attention and some techniques to control it.

This one is an attempt to integrate the extant terminological mess: it will give you many key words and a sense of despair.