The dichotomy between reason and emotion is a cultural construct, and neither grounded in neuroscience nor in logic.

1. All serious cognitive philosophy since at least Kant has supported the idea of separate faculties -- in his case, specifically the four chambers [`--` quantitative sensitivity, `-+` qualitative perceptivity, `+-` causal judgement, `++` modal reason].[1] This is also at the root of the induction vs. deduction [vs. abduction] debate, which had been raging for at least 2,200 years before Kant.

2. Even our pre-gallilean understanding of neuroscience arrived at the same conclusion, though in hilariously abstract terms: 'fast thinking' vs. 'slow thinking'. Popularized primarily by a popsci book (?) that frames it like this: "System 1" is fast, instinctive and emotional; "System 2" is slower, more deliberative, and more logical.

3. Psychologically the clearest paradigm is Freud's `- Id`/`+ Ego`/`* Super-Ego`, which is not nearly as discredited as his famous theories about parental lust would imply. If you asked your psychiatrist(/local research psychologist) about it today, they'd ofc use less medieval words for the same concepts -- roughly 'unconscious', 'cognition', and 'executive', in order.[2]

4. In cognitive science more broadly, a lot of the binaries discussed on Wikipedia's Mathematical Model article[3] are getting at this delineation.

5. Eastern philosophy has plenty of stuff on this dichotomy; in fact, I'd argue their discussion of it is predicated upon its existence and the various harmful states it can create if left unexamined. See, for example, https://en.wikipedia.org/wiki/Buddhist_logico-epistemology

On a 'logical' level, the case for a separation of at least two types of thinking seems clear: any cognitive activity can either call the 'lightweight' model to guess what's happening (and be right 95% of the time), or call the 'superheavy' model to iteratively reason through webs of beliefs.

Sure, any call probably uses more than 0.0001% of both so they're not mutually exclusive, but I think the qualitative diffence is clear.

The sentiment is true in a bit more subtle of a way, I will admit... It's hard to articulate in brief (/at all), but I've become fascinated with the way the presence of the higher faculties sometimes seems to completely transform the lower ones, unlocking whole new affordances. Like the existence of a new structure of a higher level of complexity completely changing the behavior of nearby cellular automata. If you'll excuse the anthropomorphization of abstract patterns for moment: it reminds me of how a competent leader can unlock whole new affordances for the individuals on their team without actually directly changing the individuals themselves.

Like, consider how we have have tons of both instincts & intuitions related to language use -- speaking, writing, listening, etc. Chimpanzees & bonobos share many of our incredibly advanced intuitive abilities (e.g. manual dexterity, social 'theory of mind'/empathy, etc.), but they are categorically unable to pick those up. In this way the super-ego didn't add funtionality to our species, and it didn't even really multiply it geometrically (?) -- it seems much closer to an exponentiation of capability.[4]

TL;DR: Acknowledging the fallibility and inherently-subconscious basis of reason is a noble goal, but abandoning the Idea altogether is misleading. IMHO!

[1]: http://staffweb.hkbu.edu.hk/ppp/ksp1/toc.html

[2]: https://en.wikipedia.org/wiki/Cognitive_therapy#Basis

[3]: https://en.wikipedia.org/wiki/Mathematical_model#Classificat...

[4]: Here's to hoping that this fourth rung is the last one, at least for a while! Cause if there's a fifth one, and AI discovers and produces it... yikes. Truely alien ASI would follow, rather than just ASI-as-huge-but-fundamentally-grokkable-AGI.)

The PopSci book you're thinking of is "Thinking, Fast and Slow" by Daniel Kahneman