Subitising is looking at a collection of objects and, as gestalt, assessing the number present without manual counting. It's about as basic of a System 1/2 split as you can get: children need to count manually, with each tallied increment taking effort; as we get used to seeing small collections of things in the real world, we group them together and treat the collection as atomic.

In digit grouping - e.g. presentation not as 1000000, but as 1,000,000, with the use of comma as delimiter - we again subitise chunk-by-chunk, determining the latter representation to be one million far faster than the former. We subitise straight to "two chunks of three zeroes", whereas without that convenience we often need to manually tally our million.

What is the self?

When we speak of "I", there's a bunch of targets and capabilities. For example, "I remember X", "I can Y", "I value Z". And so when we talk about the self, we combine this collection of capacities into a coherent gestalt, subitised down to one, an "I". And on thinking about this, I don't much like it.

The self that remembers has episodic, semantic, procedural memory components, each linked to but not properly contingent on the others. The removal of, say, my procedural memory of how to type isn't really, in an especially important way, a removal of "who I am". While I have many formative memories, no specific recollection is properly core to the self I would formulate.

Similarly, manipulation of any specific value retains my sense of self. If you were to induce in me some pathological murderous desires, leaving all other elements intact, the result would still in some sense be "me". At the same time, we're quite comfortable saying that, if I entered into such a state, I'm "not myself".

The boundaries are fuzzy here. There is some sense in which the above image contains "two triangles" - the outer perimeter, with a point facing upwards, and the other one being eaten by three pac-men. Neither triangle is "really there", in the sense of a contiguous shape, but when I refer to them, I imagine you have some understanding of The Thing I'm Pointing At With Words.

If I were to erase just one black pixel in the above image, I haven't really "removed" either triangle. Yet there's also some threshold of pixels I can remove beyond which there's a blank image, and neither triangle would remain present in any sense. The exact boundary of "triangle removal" is dependent on the order and number of pixels removed - it's an "I know it when I see it" kind of thing.

My "self" is like a triangle. There is a sense in which any small perturbation to me retains the "selfness" present, which when compounded will eventually eliminate it. The version of me with any specific atom removed would comfortably remain "still me". But random removal of 90% of my atoms would probably eliminate whatever me-ness remains.

I use this as a model for counterfactuals. When someone says "I could have done it differently", what they actually mean to say is, "my category of self-ness would include a version of me which took different actions". The disagreement seems almost always to focus on "could have", with the "I" being an assumption about the particular configuration of atoms composing your cells composing your brain and body and what have you. But it's the "I" that is the fuzzy bit.

In vernacular usage, we pretend "I" is a unified thing, subitising up not just disparate components of our own brain, but abstract variations on it which don't "really" exist, not seeing them as individual components but as the One Collection That Is Me. We distend our sense of self across counterfactuals, a high-dimensional set of acceptable perturbations. But then, this is obvious, in so far as "self" is something that exists across time. The "you" that started reading this sentence isn't quite the "you" that finished it, but you're comfortable with the literary convenience of saying they're "still me". Your self's smeared across time and space and dimensions that aren't real - we're weird little blobs, are humans.