Back
🚧 Work in progress Nov 18, 2026 Bas Baudoin

Automatic Cognition and Defensive Reasoning

There are different types of reactions to advice, especially unsolicited advice. Here I attempt to categorize them by ego, awareness, and the automatic cognition that comes before both.

The advisee: a shorter route

Someone tells you there is a tunnel that cuts five minutes off your walk to work. You have walked the long way for ten years. Assume they are right.

The first thing that happens is a sting: ten years, the wrong way? That sting is friction, and it shows up in everyone, however self-aware. Right behind it, before any decision is made, a reason appears: my route is probably just as fast. Nobody chose to think it. It came from automatic cognition, the thinking that runs without deliberate control. A reason that arises this way, to ease the friction rather than to weigh the advice, is a friction reason (coined).

What happens next is where the reactions split:

  1. Admission. “Ten years the long way. Damn, good tip.” The friction is felt and named, and nothing is defended.
  2. Honest decline. “Interesting, but I’m tired. Not now.” No claim about the route, so nothing false.
  3. Bare decline. “Thanks, I’d rather not get into it.” No reason given, and none owed.
  4. Sincere friction reason. “My route is just as fast,” and the speaker believes it. The reason passed through unnoticed: an unconscious misjustification.
  5. Knowing friction reason. The same sentence, but the speaker knows there is nothing behind it. A white lie that protects the ego.
  6. Vague out of it (coined). Asked “so why not take the tunnel?”, the answer becomes “well, everyone’s different, there’s a lot to consider.” Too vague to examine, which is the point.

Between 4 and 5 the words stay the same. What changes is awareness: whether the friction reason gets recognized as one. What pulls toward 5 and 6 is ego protection. Taking the tunnel means accepting ten years of the wrong route, the mechanism behind implicit acceptance denial. Ego protection can run unaware, aware, or somewhere in between: I know this is nonsense, and this is where I stop thinking about it.

One reaction is missing from the list because it is not a friction reason: “I know about the tunnel. I walk along the river for the view.” A reason that existed before the advice is a preference, not a defense.

The advisor: Raycast

Now swap seats. You show a smart, technical colleague Raycast: a launcher that opens any app by typing its first letters, with no hunting through the Dock (why that matters). Two minutes to install, almost nothing to learn, useful on day one. The answer: “Nah, my workflow works fine.” Between the lines: please stop talking about it.

From this seat the refusal looks irrational, because the cost is so low and nearly everyone values their time. An honest “I see the benefit, I just don’t care enough” would be a perfectly fine answer. It is rarely the one given. The usual answer disputes the benefit itself: a friction reason, now heard from the outside.

Unsolicited advice also carries a second message, true or not: you have been doing it wrong. Heard that way, the advice becomes a judgment of competence, and people resist pressure on their freedom to choose (psychology calls this reactance). A stronger argument leaves less room to decline without looking irrational, so it can harden the resistance instead of softening it.

The advisor brings variables too:

  • Being wrong. “Almost no cost” is the advisor’s estimate, not a measurement of someone else’s day.
  • Discussion strength. Some people argue better and longer, and some even relax by discussing when they are tired. That advantage can win a discussion that should have been lost. Losing a discussion is not the same as being wrong.

The variables

In the order they occur:

  1. Friction. The discomfort of advice that implies a past choice was wrong. The trigger for everything below.
  2. Automatic cognition. Produces a reason before any conscious decision. It produces most thoughts; in a friction situation, it produces defenses.
  3. Friction reason. Post-hoc, aimed at easing the friction, and unconscious in origin by definition. A conscious layer can recognize it afterwards. That changes what is done with it, not where it came from.
  4. Awareness. The conscious layer: whether, and how quickly, a friction reason is caught. A degree, not a switch, and it varies by situation within one person. Psychology’s word for this is metacognition.
  5. Ego protection. The pull to defend a past choice or a self-image. Unaware, aware, or half-aware.
  6. Response. What reaches the other person: admit, decline, defend sincerely, defend knowingly, or vague out of it.

The sequence is the core of the model: a friction reason is generated automatically, then recognized or not, then endorsed, revised or dropped. On this model, more awareness does not mean fewer friction reasons, only more of them caught.

The claim: NPC-like behavior

Without a sufficient awareness layer, animal impulses translate more directly into behavior, and that behavior turns automatic and rule-based: NPC-like.

The impulses behind friction are old ones, shared with animals: avoid the threat, defend the status. Awareness is the only layer that can stop them on the way out. Without it the path is short: friction, friction reason, spoken. Nothing in between asks whether the reason is true.

A short path is a predictable one. The same input produces the same output: advice that implies a mistake gets a defense, whatever the advice is about. That is how a non-player character in a game works. It follows its rules and never examines them.

The claim is about behavior, hence the like. It describes a reaction, not a person, and it runs on a spectrum: one defensive reaction says little, the same reaction across many situations says more. It is not a claim about intelligence either. A sharp mind without the awareness layer can be very good at defending, because the reasoning then works for the impulse instead of checking it.

What is defensible

Two rules, valid in both seats:

  • Honesty: don’t give a reason you know to be false.
  • Autonomy: nobody owes anyone a discussion, or a reason for declining one.

Declining is therefore never the problem. From the advisor’s seat, “I’m tired, not now” can look like avoidance, but it is a reasonable call about a limited resource. A sincere friction reason breaks neither rule either: it is an error, not a dishonesty. The problem starts when a reason is volunteered that the speaker doesn’t believe, or when a question is dodged while pretending to answer it.

For the advisor, the same rules mean: don’t use discussion strength to win, and don’t turn stamina into a demand that the other person keep going.

Limitations

  • The observer can be wrong. The advice may be bad, or the reason valid. Calling something a friction reason is a judgment, not an observation.
  • No diagnosis from the outside. Reactions 4, 5 and 6 can be word for word the same, so the model sorts one’s own reactions far better than other people’s.
  • Research status. That people sincerely give reasons that miss the real cause is well documented: Nisbett and Wilson (1977) on the limits of introspection, and the choice blindness experiments in which people explained choices they had never made. That the full reason forms before awareness is not established. “Unconscious in origin” is this model’s definition, not a finding.