There’s no such thing as “accidentally”

You knock over your coffee reaching for your phone. “Sorry, that was an accident,” you say, grabbing paper towels. But was it? Your sensorimotor system successfully navigates cups thousands of times, while your world-model neural networks understand perfectly well that laptops and coffee don’t mix. Yet somehow, at this particular moment, with this particular cup, these sophisticated systems “accidentally” failed to coordinate. Many of what we call accidents contain far more agency than we recognize. Some may be goal-directed actions hidden from conscious awareness. Others result from attention management choices that accept certain risks. Either way, they involve more agency than our outdated folk psychology admits. We mistakenly look for agency under the streetlight of conscious intentionality, trapped in what Daniel Dennett calls the Cartesian Theater.

The Cartesian Theater Is Closed

When we think about intentional action, we imagine a little person in our heads — somewhere behind our eyes, watching our inner movie, deliberately choosing each action. When this executive is absent, we call the resulting actions “accidental.”

But this Cartesian Theater is an illusion. There’s no unified soul-like Self sitting in a control room pulling levers. Much of our sophisticated behavior is the output of what Minsky called a “Society of Mind” — not a single commander but a coalition of semi-autonomous systems, each with its own goals, much of it happening without conscious oversight at all. If agentic action doesn’t require a conscious overseer, then what distinguishes the coffee spill from any other action? The distinction turns on something other than conscious intention: the capacity to model consequences.

Cognitive Light Cones

Consider two utterances: calling someone the wrong name in conversation versus a Tourette’s outburst. Both involve unexpected vocalizations, both can be socially awkward. Yet one feels like a mistake while the other feels like something happening to you. The difference lies in what biologist Michael Levin calls “cognitive light cones” — the spatio-temporal boundary of events that a system can measure, model, and try to affect. This boundary defines the scale and limits of cognition. The neural circuits producing Tourette’s vocalizations have a minimal cognitive light cone; they cannot measure social reactions, model reputational consequences, or affect outcomes beyond immediate tension release. They operate below the level where social consequences exist within their computational boundary. Your utterance-producing system, the non-Tourette’s version, is radically different. When you call someone the wrong name, it’s not just producing sounds but potentially drawing on vast networks that track social relationships, emotional associations, conversation goals, and interpersonal dynamics. This system has a much larger cognitive light cone. When this sophisticated system produces the “wrong” name, should we always dismiss it as mere accident? The capacity for consequence modeling was there — maybe it just happened to produce an outcome that surprised the conscious Self.

The Social Tell

Our social reality is a strong contender for a domain where some of our goals are hidden from the imaginary Cartesian Observer (our conscious Self). Our social cognition systems evolved over millions of years to navigate complex human dynamics. They track status hierarchies, coalition politics, mating opportunities, and reputation effects — all without conscious awareness. Notice how some people “unintentionally” steer conversations toward their accomplishments and much rarer toward their failures. Should we conclude that their sophisticated social modeling systems just randomly surface flattering anecdotes? Or are they doing exactly what they evolved to do: manage status while maintaining the appearance of casual conversation? Even when something is said intentionally, its true purpose can remain hidden from the speaker. Consider someone who tells a friend “you’d be so much happier if you just left him.” If the friend agrees, it was heartfelt advice. If she bristles, it was “just a joke, obviously.” The statement exists in social superposition: its meaning collapses based on the other person’s reaction, not on the speaker’s “true” intention. Less controversially, isn’t it plausible that there are multiple reasons for saying a particular thing, more than we were consciously aware of before speaking? These are reasons that We may or may not endorse but nevertheless recognize after the fact that they caused our speech.

“I was just distracted”

“I was just distracted” can be simultaneously a courageous admission and a vacant excuse. Distraction isn’t something that happens to us. It’s something we do. Our attention management system itself operates within a cognitive light cone: it can measure attention demands, model consequences of divided focus, and directs information processing. We model the consequences of divided attention, which is why we usually put the coffee down before reaching for the phone. When we don’t, we may not be consciously aware of it but we’ve made an implicit calculation: the risk of spillage is worth the efficiency of multitasking. The system that allocates attention has goals (efficiency, stimulation) and models consequences (spillage probability). The spill might represent a gamble that didn’t pay off, or, more intriguingly, it might serve hidden goals like regulating caffeine-induced jitters when anxiety levels are rising. Either way, it’s not a random event.

But I have ADHD / But I have anger issues

Whether we spill the coffee or not depends on our cognitive light cone at that moment. This in turn is not genetically predetermined (at least not entirely) but something we actively shape. Every time we practice mindfulness, we expand the boundary of what we can measure, model, and affect. Conversely, every time we multitask carelessly or scroll mindlessly, we reinforce habits that shrink this boundary.

Consider someone who repeatedly responds to anger with physical violence. Each time they choose this response, they’re not just acting in the moment — they’re building a neural highway that shrinks their cognitive light cone. What they can measure, model, and affect during anger becomes increasingly narrow. Ten years later, when they lash out automatically, that action is less intentional in the moment precisely because they’ve trained themselves into truly being incapable of responding any differently. They’ve authored their own diminished agency. The “accident” of uncontrolled violence is really thousands of intentional moments incrementally cementing the habitual response.

Sure I agree, so what?

What changes when we recognize accidents as more purposeful? First, it expands your sense of agency. That coffee spill isn’t proof of clumsiness — it’s evidence of a complex system making complex tradeoffs. You’re not a passive victim of accidents but an active participant in creating them. Second, it suggests mindfulness isn’t just spiritual practice but agency expansion. When you practice attention management, you’re literally growing your cognitive light cone — expanding the spatio-temporal boundary of events you can measure, model, and affect. You begin noticing the subtle tension that precedes snapping at someone, the fatigue that makes you clumsy, the restlessness driving impulsive reaches for your phone. These internal states were always there, but now they fall within your computational boundary — you can track their patterns, predict their consequences, and intervene before they cascade into “accidents.” Where you once had only automatic reactions, you now have choice points. Third, it invites curiosity about your errors. Instead of dismissing mistakes as random, you can ask: what goal might this serve? What is my social modeling system trying to accomplish with this “slip”? This isn’t psychoanalytic mysticism — it’s recognizing that your brain is good at what it does, even when You are not watching.

Beyond Accidents

The title “There’s No Such Thing as Accidentally” is deliberately provocative. Sometimes we just spilled our coffee, it’s not that deep. But sometimes, the purposefulness hidden in our “accidents” sheds light onto a post-Cartesian Theater view of psychological reality: we’re not all-powerful, all-knowing conscious Gods controlling meat puppets, but we’re not passive victims of random mechanistic neural firings either. We’re something more nuanced: distributed systems with overlapping, sometimes conflicting goals, navigating the world through consequence modeling that exceeds our conscious awareness. Our accidents are sometimes intentional, We just weren’t invited to the planning meeting.

The next time you spill coffee, pause before claiming it was accidental. Your motor system knew the risks. Your attention system chose its priorities. Your anxiety-regulation system might even have had reasons. You are composed of multitudes, and they’re paying attention — just not to what you think. The dualist Cartesian Theater is closed, but the show goes on.