Thread by @j_theriault

We have a new paper accepted at Physics of Life Reviews! We outline a model of social pressure, showing how it may be individually adaptive to conform to others’ expectations, with radical implications for accounts of social cognition and prosocial behavior. To foreground a few:

We develop a novel explanation for status quo biases.

We suggest that prosocial behavior may be motivated (in part) by short-term predictability (rather than long-term reciprocity).

We outline how historical contingencies and social constructs may interface with brain biology.

I’ve outlined the major line of argument in the tweets below, but for anyone interested I’d encourage you to check out the full paper, available at Physics of Life Review, here:
sciencedirect.com/science/articl…
Or the free preprint here:
psyarxiv.com/x5rbs/

Influence carries many meanings. We may be influenced to copy when we assume others know something we don’t (informational influence). We may be influenced by seeking positive/negative reputation (reputation-seeking). We may be influenced by our moral values (moral obligation).

We are interested in a slightly different sort of influence, often overlooked (like water to fish, to borrow from DFW). We were interested in the influence of others’ expectations, where those expectations exert an influence in and of themselves.

We call the subjective experience of this influence a sense of should. We mean “should” in the normative, but non-moral sense—a sense of should may or may not oppose your moral values, but when people expect a behavior from you, you may feel a social pressure to do it.

To give an example, remember that you are an animal, then think about why an animal would squirm in discomfort during a lecture, waiting to pee, when it could just stand up and walk away. Or, if you are male, would you feel some anxiety wearing nail polish for the first time?

These may seem like trivial examples, but this is actually quite important, as reputation and social sanctions are often used to explain why people conform to norms. But many norms are clearly arbitrary, and are unlikely to be punished. Why follow them?

Also note that behaviors are multiply determined—you may not feel anxious about flouting gender norms; or, your boss may fire you for sneaking out of a meeting. But, often reputational threats are unlikely, and we offer a general account of what motivates behavior in those cases.

In the paper, we outline: (1) a foundation for our framework in neuroscience and metabolics; (2) a simple model of how a sense of should operates, is experienced, develops, and intersects with mental inference; (3) its implications for econ, social psych, and evolutionary theory.

We grounded our framework in metabolism. “At its biological core, life is a game of turning energy into offspring @HermanPontzer”. Metabolics is often overlooked in psych—WEIRD culture may put food concerns (for the well-off) into the background—but it is evolutionarily critical.

The brain is also energetically expensive, and accounts for ~20% of resting energy consumption in humans (~9% in chimps; ~ 5% in rats). Recent accounts suggest the brain encodes info efficiently by hierarchically predicting sensory signals, then filters what was predictable

Predictive coding is at the core of information theory, which makes signal compression possible (e.g. in .jpeg images or your cell phone). We add that this compression may regulate the metabolic costs of neuronal signaling as well (which are ~75% of ATP costs in grey matter).

If this is right, then predictable environments are metabolically efficient environments. The metabolic costs of bottom-up signaling may be eased when the environment is predictable, with the caveat that predictability is balanced with other metabolic/fitness concerns.

From that foundation, the model becomes social when you consider this:
By making yourself predictable to others, others can be made (and kept) predictable for you.

A social environment can be kept predictable by inferring agent’s expectations and matching your behavior to them.

In other words, if brains make predictions, encode prediction error, and change behavior to respond to uncertainty—then YOUR social environment can be kept predictable by inferring what OTHERS expect, and conforming.

We outline a strategy for creating and exploiting predictability in a social environment. Of course, people do not ONLY try to minimize prediction error. People explore and break taboos. But by exploring, they learn how others react, making more accurate predictions in the future

With this model in place, we outline why a sense of should *feels* different than simply wanting to gain status or avoid sanction. We suggest that the feeling stems from anticipated affect—i.e. you learn that violating others expectations generates uncertainty and arousal.

If a sense of should stems from anticipated arousal, then it may be reinforced without intentional punishment. i.e. Others don’t need to punish norm deviations, they just need to react to surprises. Your affective reaction to changes in their behavior conditions your conformity.

We then sketch a developmental trajectory, as a sense of should may be experienced differently across the lifespan. As your predictions become more accurate, the cause of prediction error and arousal (i.e. violating others’ expectations) may become more learnable.

Next, we outline how this relates to mental inference. As our model is centered on the relationship between others’ predictions and your behavior, we suggest that this relationship can be leveraged to fine-tune mental inference (in combination with prior knowledge).

The implication of this account is that it will be difficult to make accurate inferences about others’ minds without interacting with them, as precise mental inference depends on deploying and testing hypotheses about what others expect you to do.

Finally, we underscore that conforming to others’ expectations should be understood as a form of influence in itself. Conforming exerts a stabilizing influence—it won’t make others more likely to do something specific, but it regulates their behavior by keeping it predictable.

Now, implications!

First we outline how our model could be improved. We avoided talking about predictive precision, where is standard in predictive processing models. A full account should include these. We also review primate work, where a similar but less sophisticated strategy is deployed.

On status quo biases, if (individually) conforming makes the (collective) social environment predictable, then violating shared expectations (norms) will disrupt the environment for everyone. Enforcing a status quo may regulate predictability in a collective social environment.

However, predictability (and metabolic efficiency) is only one facet of evol. fitness. What if you are personally harmed by the status quo? Unless it is intolerable, you may feel conflicted. Things could improve by challenging it, but things are predictable if it is maintained.

In other words, the more predictable a social environment, the more each individual stands to lose by disrupting it. This is consistent with work by @SimonDeDeo: Wiki editors use disruptive “revert” actions in bursts, but decrease use as things stabilize.
journals.plos.org/plosone/articl…

Next, we outline how our model relates to communication. We focused on conforming to others’ expectations, but if you can influence those expectations you might regulate the environment via another route: guiding others to accurately predict your actions.

In behavioral economics, it is usually assumed that selfishness provides a short-term benefit and cooperators benefit long-term. We suggest that conforming to others’ expectations provides an ultra short-term benefit. The reward is a predictable environment, made RIGHT NOW.

Our model also ends up looking very similar to some models developed to explain “guilt aversion”, although our explanation is not specific to guilt, and explains the involvement of affect in decision making more generally.

Maybe most exciting to me: our model makes clear how social constructs and historical contingencies interact with individual psychology and brain biology. Critically, for a sense of should, conforming to an expectation is adaptive REGARDLESS of its content.

Note: conforming might harm or help for other content-related reasons—e.g. ritual self-harm. BUT, for the purposes of maintaining a predictable environment (which, we argue, has fitness consequences) it ONLY matters that you conform to expectations, no matter how arbitrary.

So, if a behavior becomes expected by historical accident (eg. marriage customs; see below), then an arbitrary norm may become self-reinforcing. This means that observed human behaviors cannot be assumed to be adaptive. Psychology must incorporate history
science.sciencemag.org/content/366/64…

Finally, we connect our model with its implications for motivated cognition. If you feel compelled to behave as others expect, and others can accurately infer your beliefs, you may feel compelled to adopt or dispel certain your own beliefs, in accordance with that pressure.

Thank you for reading! This paper has been a labor of love over the last couple of years, so if you’ve met me I’m sure I have talked your ear off about it. Thank you to everyone who has listened along the way, and to my fantastic advisors @LFeldmanBarrett and @LianeLeeYoung!

In the next few months, we will also receive and respond to comments, to be published in Physics of Life Reviews. So I look forward to continuing the conversation and elaborating on these ideas with the help of others!