The eye that
predicts, not reacts.

Chapter I

The dragonfly as predator —
95% interception

Dragonflies are the most successful aerial predators in the animal kingdom by kill rate. Unlike most predators that react to prey location, the dragonfly computes where prey will be and flies to that point — an internal prediction loop running faster than any conventional sensorimotor reflex.

The neural substrate responsible is a small population of neurons called target-selective descending neurons (TSDNs), identified through the work of Steven Wiederman and others. These neurons fire selectively to small moving objects — particularly objects moving against a complex, textured background — and project directly to flight motor circuits. The dragonfly does not deliberate. The prediction is the action.

Critically, TSDNs encode the future position of a target, not its current one. Studies of dragonfly interception trajectories show flight paths consistent with predictive pursuit (constant-bearing decreasing-range geometry) rather than reactive tracking. The neurons fire ahead of the target's arrival.

Chapter II

The Hassenstein-Reichardt detector —
motion from correlation

In 1956, Bernhard Hassenstein and Werner Reichardt described a model of directional motion detection derived purely from behavioral experiments on walking beetles. The model contains no optics, no biology — only logic: if a stimulus activates two adjacent detectors in sequence, and the delayed output of the first is multiplied with the instantaneous output of the second, the result is direction-selective.

  Photoreceptor A          Photoreceptor B
        │                        │
        ├──── delay (τ) ─────────┤
        │                        │
        └──── × ─────────────────┘   ← multiplication
                    │
              Direction signal

  Mirror-symmetric subunit (A←B) subtracted:
        └──── × (null direction) ────┘
                    │
              Net directional output

  Key operation: temporal correlation via delay + multiply
        

The model predicted, from behavior alone, that motion computation would require: a temporal delay between adjacent channels, a nonlinear interaction (multiplication), and subtraction of opponent directions. All three were subsequently confirmed in electrophysiology — decades later.

What made the model remarkable was its predictive power across stimulus types. Gratings, apparent motion sequences, stochastic flicker — the Hassenstein-Reichardt correlator reproduced recorded neural responses quantitatively across all of them. It remained the governing framework for insect motion detection for nearly 70 years.

Chapter III

T4 and T5 neurons —
the circuit identified

The cellular elements of the Reichardt detector were identified in Drosophila over the last decade, principally through work in Axel Borst's group at the Max Planck Institute of Neurobiology and connectomics efforts at Janelia Research Campus.

The fly optic lobe has a columnar organisation: photoreceptors → lamina → medulla → lobula → lobula plate, each column processing one facet of the compound eye. Within this structure, T4 and T5 neurons are the primary motion-sensing cells.

• T4 neurons respond selectively to ON edges (light increments) moving in their preferred direction
• T5 neurons respond selectively to OFF edges (light decrements) — a parallel channel, like the mammalian retina's ON/OFF split
• Both come in four subtypes, each tuned to one of four cardinal directions: right, left, up, down
• Each subtype projects to a distinct layer of the lobula plate, creating a four-layer directional map visible under two-photon imaging

Downstream of T4/T5, lobula plate interneurons perform the subtraction stage — they receive excitatory input from one directional layer and inhibit the adjacent opposing layer. This implements the null-direction suppression that sharpens the tuning curve beyond what either mechanism achieves alone.

Chapter IV

Dual mechanism —
preferred-direction enhancement and null-direction suppression

For decades, two competing theories of directional selectivity existed in parallel. Hassenstein and Reichardt proposed that motion detection works through preferred-direction enhancement: the delayed signal from one photoreceptor is multiplied with the undelayed signal from the next, amplifying responses when motion crosses in the preferred direction.

Barlow and Levick proposed instead null-direction suppression: a delayed inhibitory signal from the null direction vetoes the response before it can propagate.

Both theories were right. Research using apparent-motion stimuli at individual facets of the fly eye showed that the answer depends on where within the receptive field you probe: the leading edge shows preferred-direction enhancement; the trailing edge shows null-direction suppression; the center shows both simultaneously.

The biophysical substrate of this dual mechanism was identified through receptor mapping on T4 cell dendrites. GluCl-α receptors (glutamate-gated chloride channels — inhibitory) are concentrated at the dendritic tips, where Mi9 input arrives. Nicotinic acetylcholine receptors cluster in the center, where Mi1 input arrives. GABA-A receptors sit at the base, driven by Mi4.

When a bright edge crosses the receptive field in the preferred direction: Mi9 (an OFF cell) is silenced — its chloride channel closes — causing the dendritic tip's input resistance to rise, which amplifies the central excitatory Mi1 signal. Motion in the null direction causes the opposite sequence, suppressing the response. The geometry of the dendrite performs the computation. No learning. No weights. Fixed physics.

Chapter V

Why this matters for
event-based silicon

Dynamic Vision Sensors (DVS cameras) output asynchronous spike events — one per pixel, per polarity change, timestamped to the microsecond. This is functionally equivalent to the output of retinal photoreceptors in the fly: sparse, temporally precise, polarity-coded signals rather than continuous luminance values.

The biological circuit that evolved to process exactly this type of input — and to extract direction and future position from it in nanoseconds — is the Hassenstein-Reichardt pipeline. LIBELLULA's architecture maps that circuit onto synthesizable RTL, preserving the essential structure: spatial delay, temporal correlation, direction-selective multiplication, confidence gating, and predictive integration.

• The delay lattice implements the spatial offset between adjacent photoreceptor columns
• The LIF tile implements the temporal filtering and integration that laminar and medullary neurons perform
• The Reichardt detector implements the multiplication and subtraction identified in T4/T5 circuits
• The α–β predictor implements the predictive integration that dragonfly TSDNs perform — outputting a lead-point coordinate, not a current position

The result is deterministic behavior at every stage. The output of each module is traceable from input event to predicted coordinate without any learned parameters, stochastic inference, or black-box processing. This is what biological neural circuits actually do — and it is the design target for LIBELLULA.