If internal inattentional blindness is real, people could plausibly learn to see more clearly into their own mind through attentional training, which would have huge implications both for our scientific understanding of the mind and for interventions to improve well-being.

In this paper, I ground the idea in theories of perception and internal attention, synthesize existing indirect evidence for it, and chart a roadmap for how to test it directly.

This idea of "internal inattentional blindness" is implicitly assumed in much of clinical psychology and public discourse, but it has not been given precise theoretical or empirical treatment.

In other words, many high-level mental processes may not be *intrinsically* or permanently unconscious. Rather, people may be experiencing an internal analogue of inattentional blindness -- they fail to see what's going on in their mind because they're not paying attention.

A widespread view in psychology is that most cognitive processes are unconscious. In a new paper, I argue that many of these processes evade awareness for the same reason the "invisible gorilla" did: People fail to pay attention to them. 🧵.

These results challenge the idea that most decision processes are unconscious, and instead suggest that – at least in simple choices – people are often aware of how they’re choosing. Next, we’re testing awareness in more realistic contexts, and how that awareness can be improved!.

On the other hand, awareness varied enormously across participants, suggesting that some people may be better at introspecting on their choice processes than others. Awareness was not predicted by participants’ self-reported confidence or self-reported introspective ability.

We also recruited “observers” who were matched with the original “actor” participants, shown their choices, and asked to predict the actors’ choice process. Actors were more accurate than observers, suggesting that their accuracy came from some kind of first-person introspection.

Before the study, we asked decision scientists from SJDM to predict participants’ accuracy. Participants were *much* more accurate than these experts predicted; experts thought people would only show an r of 0.44. (Thanks to everyone at @SJDM_Tweets who participated!)

The main result: Participants’ self-reports were often highly correlated with their actual choice processes. One study found an average correlation of 0.8 between self-reported attribute weights and the best-fit weights from the computational model.

We then fit computational models to their choices to identify their actual attribute weights & choice strategy. Our models accounted for diverse choice strategies (e.g., rather than assume everyone computed full expected values, we accounted for people using heuristics).

Participants made value-guided choices (e.g., between homes to rent) that varied on many attributes at once (size, kitchen quality, etc.), and then reported how they think they made those choices: how much weight they placed on each attribute, and which heuristics they used.

Are we “strangers to ourselves”? Classic theories say people have limited insight into how they decide. Our new paper at @NatureComms challenges this view. With @carlsonr_, @hedykober, & @mollycrockett. 🧵.

If you're applying to psych PhD programs this year and want feedback on your application statement, check out ASFP! Students from marginalized backgrounds and those without access to informed mentorship are particularly encouraged to apply:.

Our task provides a flexible, generative paradigm for further investigating the extent and nature of introspective access to choice processes. (8/8).

These results push back against the notion that we are “strangers to ourselves”, and suggest instead that people can sometimes introspect directly on the mental processes underlying their choices. (7/8).

They were also substantially more accurate than informed third-party observers (who were shown all their choices), even when incentivizing those observers for accuracy. This first-person advantage suggests that participants' accuracy comes in part from introspection. (6/8).

Participants were surprisingly accurate: Their reported attribute weights correlated with the model-estimated weights at r = 0.5 to 0.85, and they were 2-3x likelier than chance to report exactly their true choice method (including which heuristics they used). (5/8).

Finally, participants reported how they think they made those choices, and we compared their self-reports to the modeling results to quantify each individual's accuracy about their choice process. (4/8).

Then, for each participant, we used computational modeling to identify (a) the decision weight they placed on each attribute, and (b) the method they used to come to a choice – specifically, whether they used full utility-maximization vs. various simplifying heuristics. (3/8).