Vehicle 9
Shapes
Shape becomes less mysterious once it is treated as a sequence of sensor changes over a scan rather than as a fully formed picture delivered all at once.
9 min read
The three warm points act like a crude outline. What matters is the sequence of encounters the body stitches into a single path.
DemonstrationA shape can begin as an ordered sequence of encounters. The body never sees a finished form all at once; it accumulates one by moving through it.
Shape perception often gets described as if the whole form were simply present to the system at once. Braitenberg’s framing points in a more incremental direction. A shape can be recognized through the way a sensor signal unfolds over time.
This matters because it ties form back to interaction. The signature of a square, a circle, or a triangle is not just the figure itself but the temporal pattern produced by scanning across it.
Key idea
A shape can be encoded as a changing relation over time. Recognition begins when the system becomes sensitive to the signature of that unfolding.
The signature is often enough
If a sensor sweep produces a distinctive pattern, the mechanism does not need a high-resolution image to differentiate forms. It only needs to become sensitive to the profile generated by the scan.
A shape can be represented by how sensor input changes across a scan, not just by a static picture. The right panel shows the signature that unfolds over the sweep.
Form is learned as a relation
The broader lesson is that perception is rarely a single instant. It is often the stabilization of a relation between body, sensor, and world over time. Shapes are one clear example of that rule.
A shape can be represented by how sensor input changes across a scan, not just by a static picture. The right panel shows the signature that unfolds over the sweep.