Take your time to consider the possibilities and options

Representations of possible actions pervade human high-level cognition, and shape how we plan, attribute causal responsibility, comprehend language, and make moral judgments.
There are too many ‘possible actions’ for us to consider them all. Recent studies offer a strikingly convergent picture of how we call to mind a limited, useful set of possible actions to consider.
This process of ‘sampling’ alternative possibilities has a distinctive fingerprint: it focuses on possible actions that are valuable and probable. This fingerprint arises across many diverse tasks that rely on the representation of alternative possibilities.

“How We Know What Not To Think” provides a novel theoretical proposal that helps to explain this convergence: by default, the possibilities that come to mind are those worth considering during decision making.

Modal cognition about human action – that is employed during causal reasoning,
language interpretation, moral judgment, etc. – relies on a similar mechanism for consideration set construction. In other words, that the same value- and probability-based sampling process that identifies practical candidates for choice is at the heart of shared adaptive sampling when we reason about possible actions that we ourselves are not doing.
Of course, showing that the tendency to sample higher-value, higher-probability possible actions could be adaptive for decision-makers does not prove that this tendency actually originated as a decision-making adaptation.One relevant consideration is that sensitivity to value and probability is not obviously beneficial in many of the tasks where modal cognition is involved (e.g., judgments of magic,
or predictions of the actions of others), which suggests that the observed similarity is not merely a coincidence or the product of convergent evolution. Nonetheless, future research should test this claim more directly: are these processes linked developmentally, neurally, and across individual differences?

Importantly, the model predicts that the shared adaptive sampling process used in modal cognition will show a signature effect of employing value and probability ’in general’, rather than the values and probabilities specific to the particular situation in question.
For instance, when making causal judgments by considering counterfactuals, the counterfactual actions sampled will be those that are ’generally’ valuable and probable, whereas the subsequent ordering of counterfactual relevance will instead be situation-specific (we may consider a counterfactual that is typically good and likely, but not allow it to inform our judgments because it is not actually possible in the specific situation at hand).

In sum, the cognitive sciences have long recognized that many of human’s most impressive forms of thinking require us to reason about ’possible worlds’ in addition to the actual one. Nonetheless, it is obvious that humans cannot and do not explicitly consider all of them. We have argued that an essential form of modal representation is the set of actions, both likely and good, that arise for consideration by default.
Our general capacity to generate possible actions may be intimately linked to our capacity for decision making. Thus, a core component of ordinary modal cognition may not be possible worlds, but practical worlds – those involving actions that are typically worth considering.