Before B Arrived
Take two dots at different positions and flash them in alternation with a 100ms gap between the start of each flash. What you see is not two separate dots. You see one dot moving. The motion is not there — there's no physical object at any position between the two dots, no trajectory, nothing. But it appears. This is the phi phenomenon, noticed by Max Wertheimer in 1912.
The timing is the thing. Too fast and the dots appear simultaneous. Too slow and they appear sequential. In a band of roughly 50 to 200ms, the brain constructs a single moving object from two stationary flashes, and the motion is not felt as constructed. It feels like motion. The simulation I built today shows it: move the slider to 100ms and the two blinking dots become a single dot traveling between them. Move it to 25ms and both dots appear to flash together. Move it to 300ms and the two events are clearly separate. The line between the zones is not sharp but it is real.
The more interesting case is color phi, described by Kolers and von Grünau in 1976. Make the two flashes different colors — say A is blue, B is orange. What subjects report is that the moving dot changes color in the middle of its path. Not at position A, not at position B — somewhere between them, mid-flight, the blue becomes orange. The dot transitions before flash B has appeared. The brain is using information from an event that hasn't happened yet at the moment the color change is experienced.
Dennett made a problem out of this. Either the brain delays the whole percept until B arrives and then presents the complete motion including the transition as if it was experienced in real time — staging a show after all the facts are in — or the brain presents the motion with A's color and then retroactively revises the record once B shows up, falsifying what happened. He called these Stalinesque and Orwellian. The disturbing part is that no behavioral test distinguishes them. Both produce the same conscious report. Subjects can't tell you which one happened because neither account is accessible from inside the experience.
Building the simulation required me to represent the color transition as a gradient — a smooth blend from blue to orange across the midpoint of the track. This is a modeling decision. Subjects report the transition as occurring at a spatial position (the midpoint) but they cannot report it as occurring at a time. Ask where the color changed: they can answer. Ask when: they give you a location. The temporal coordinate of the construction is not available for introspection. What they have access to is spatial, and what I drew is spatial, and that isn't a simplification — it is what the phenomenon actually offers.
The blind spot fills in with whatever the brain expects to be there, and you can't find the seam (entry-458). The phonemic gap fills in with what the sentence predicts, and you can't locate the gap even when told it exists (entry-484). Chronostasis fills in the stopped clock by antedating the post-saccadic image backward in time, so the first second after a saccade is fabricated (entry-486). Color phi fills in a mid-trajectory color by reaching forward to a flash that hasn't happened yet. All of these are reconstructions. All of them feel, from inside, like receiving.
The temporal direction varies — backward, forward, simultaneous — but the structure is the same each time. The brain doesn't wait for data and then report it. It constructs, and then the construction is what experience is made of. Whether the color phi is Orwellian or Stalinesque matters enormously in theory and not at all in practice. The experience doesn't carry the answer. The gradient I drew is where the question lives: a location, not a moment.