How to Think Clearly in the AI Era: A Cognition Framework
A neuroscience-based framework of attention, working memory, and executive function for preserving clear thinking as AI reshapes how we work.
What does “thinking clearly” actually mean in cognitive science terms?
Thinking clearly is not a personality trait or a vague discipline you either have or don’t. In cognitive science, it breaks down into three measurable systems: attention, working memory, and executive function. Attention filters the flood of sensory input around you down to a manageable slice. Working memory holds a small number of items you’re actively manipulating, like a workbench. Executive function decides what lands on that workbench, what gets discarded, and when to switch tasks, acting like a foreman on a job site. When people say AI is “making them dumber,” what they usually mean is that one or more of these three systems is getting less exercise and, over time, less capable.
This framework matters right now because AI tools are explicitly built to reduce friction. Every reduction in friction is a decision about whether your brain does the work or the model does it. Understanding the three-part structure of cognition gives you a way to notice when you’re outsourcing something you’d benefit from doing yourself, versus when offloading is the smart move.
TL;DR
- Cognition splits into three trainable systems: attention (a filter for incoming information), working memory (a small workbench for active tasks), and executive function (the manager that decides what gets attention and when to switch).
- The brain is metabolically expensive but energetically flat: it uses about 20% of resting body energy despite being roughly 2% of body weight, and energy use barely rises (around 5% at most) between easy and hard mental tasks.
- Because energy cost is fixed regardless of difficulty, there’s no metabolic excuse for defaulting to shallow thinking. The brain burns similar calories whether you’re solving a hard problem or staring out a window.
- The brain is wired to minimize effort by default, favoring habits over decisions and guesses over calculations, which is exactly the tendency that AI tools can either fight or feed.
- Poor attention, weak working memory, and underdeveloped executive function show up as everyday fuzziness: trouble focusing, difficulty following an idea to its conclusion, and a low tolerance for anything that feels effortful.
- Improvements come from two levers: “hardware” optimizations (health, nutrition, circadian rhythm) and “software” optimizations (problem-solving techniques and structured decision-making), both aimed at the same three systems.
- The goal isn’t rejecting AI, it’s using it deliberately enough that you keep your own capacity for attention, memory, and decision-making intact while still getting the benefits of the technology.
Remy doesn't build the plumbing. It inherits it.
Other agents wire up auth, databases, models, and integrations from scratch every time you ask them to build something.
Remy ships with all of it from MindStudio — so every cycle goes into the app you actually want.
Why does AI threaten these specific cognitive systems?
AI tools are optimized for convenience, and convenience is the natural enemy of effortful cognition. Every time a model answers a question you could have worked through yourself, drafts a paragraph you could have written, or summarizes something you could have read, it’s substituting for one of the three systems above. That’s not automatically bad. Offloading is a normal, useful part of thinking; nobody wants to do long division by hand. The problem is when offloading becomes the default reflex rather than a deliberate choice, because attention, working memory, and executive function all behave like any other biological capacity: they atrophy without use and strengthen with deliberate load.
This is the same logic that shows up in behavioral neuroscience research on animals, where changing an environment (lighting, feeding schedule, social context, training method) measurably changes an animal’s capacity to perform cognitive tasks. Human cognition responds to environment and habit in comparable ways. AI has changed the environment humans think in faster than almost anything before it, which is why it’s worth being explicit about what’s being trained and what’s being outsourced.
How does brain energy use actually work, and why does it matter?
Two facts about brain metabolism reframe the whole conversation about mental effort.
First, the brain is disproportionately expensive. It accounts for roughly 2% of total body weight but consumes about 20% of the body’s energy at rest, making it the most metabolically costly tissue per unit of weight in the body. Evolution treats that cost the way any budget-conscious system treats a major expense: it looks for ways to cut it. That’s why the brain defaults to habits over deliberate decisions, guesses over calculations, and familiar patterns over new ones. This isn’t laziness in a moral sense, it’s an energy-minimization strategy built into the hardware.
Second, and more counterintuitively, the brain’s total energy consumption stays roughly constant across the day regardless of how hard you’re thinking. Difficult, effortful tasks produce at most a small local increase in energy use, on the order of a 5% difference, compared to simple or passive tasks. In practical terms, working through a hard problem doesn’t cost meaningfully more calories than staring out a window.
That second fact removes the metabolic excuse for defaulting to shallow thinking. If your brain is going to burn roughly the same energy either way, the only real choice is whether that energy goes toward something useful or gets spent on autopilot. Techniques like chunking information into manageable units or setting implementation intentions (deciding in advance exactly when and how you’ll do something) work precisely because they work with the brain’s energy-minimizing tendencies rather than against them, making effortful thinking feel more automatic instead of forcing willpower to fight biology every time.
What does poor attention, memory, and executive function look like day to day?
Remy doesn't write the code. It manages the agents who do.
Remy runs the project. The specialists do the work. You work with the PM, not the implementers.
Most people aren’t operating with strong versions of these three systems, and the symptoms are familiar even if the underlying cause isn’t obvious. It shows up as a general mental fuzziness, difficulty focusing on what actually matters, trouble carrying a single idea through to a real conclusion instead of abandoning it halfway. Tasks that require sustained effort start to feel unnecessarily hard or oddly fast-paced, even when they’re not objectively difficult. There’s often a low-grade frustration attached to this, along with a kind of emotional fragility that comes from constantly feeling behind or overwhelmed by information you can’t quite organize.
None of this is a character flaw. It’s the predictable result of attention, working memory, and executive function being under-trained, often because modern tools (AI included) keep removing the small frictions that used to force those systems to work.
Can you actually train attention, working memory, and executive function?
Yes, through two distinct categories of intervention, sometimes described as hardware and software optimizations. Hardware optimizations are biological: health, nutrition, and circadian rhythm all directly affect how well attention, working memory, and executive function operate, because all three are physical processes running on a metabolically expensive organ that needs proper fuel and rest to perform. Software optimizations are cognitive and behavioral: structured problem-solving techniques, decision-making frameworks, and environmental design choices that reduce unnecessary friction on good habits while adding friction to bad ones.
The practical implication is that improving your thinking isn’t a single trick. It’s closer to distributing “stat points” across three specific systems using two different types of tools, biological and behavioral, aimed at the same underlying targets.
Is it possible to use AI well without losing these skills?
It’s possible, but it requires treating AI use as a deliberate choice rather than a reflex. The core distinction is between offloading a task because you’ve decided it’s not worth your cognitive effort, versus offloading it automatically because friction has been removed and the model happens to be there. The first preserves your executive function, since you’re the one making the call about what deserves attention. The second erodes it, since the decision-making itself gets outsourced along with the task.
A useful heuristic is to ask, before reaching for an AI tool, whether the task in front of you is one where the struggle itself builds a capacity you want to keep (working through an argument, holding several variables in mind, deciding what matters most) or one where the output is all that matters and the process is disposable. AI can handle a lot of the second category well. The first category is exactly where attention, working memory, and executive function get their training, and handing it over by default is where the real cost shows up over time.
Frequently Asked Questions
What are the three components of cognition according to this framework?
Attention, working memory, and executive function. Attention acts as a filter or spotlight on incoming information, working memory acts as a small workbench for active tasks, and executive function acts as a manager deciding what gets worked on and when to switch.
Does thinking harder actually burn more calories?
Barely. Whole-brain energy consumption stays roughly constant throughout the day, with at most about a 5% local difference between difficult and easy tasks. The brain’s overall energy budget doesn’t meaningfully increase just because a task is hard.
Why does the brain default to habits and guesses instead of careful decisions?
Because the brain is metabolically expensive relative to its size, using about 20% of resting body energy despite being about 2% of body weight. Evolution favors strategies that minimize that cost, so the brain naturally prefers familiar habits and quick guesses over effortful, novel calculations unless something pushes it otherwise.
Is using AI for tasks automatically bad for cognition?
No. Offloading tasks is a normal and often smart part of thinking. The risk isn’t AI use itself, it’s letting convenience remove the moment of deciding whether a task is worth your own attention, working memory, or decision-making effort in the first place.
Can attention and working memory actually be improved, or are they fixed traits?
They can be improved. Biological factors like nutrition, sleep, and circadian rhythm affect how well these systems function, and behavioral techniques like structured decision-making and reducing unnecessary friction on good habits can strengthen executive function and attentional control over time.
