Predictive coding and oscillations underlie the optomotor response in distant insect lineages

This study demonstrates that the optomotor response in distantly related insects (ants and earwigs) is not a simple stimulus-driven reflex but a complex, stochastic behavior generated by an ancestral closed-loop system integrating predictive coding and internal oscillators to minimize optic flow prediction errors.

The Gist

Imagine you are walking down a hallway while wearing a pair of VR goggles. Suddenly, the walls start spinning around you. Your brain screams, "Whoa, I'm spinning!" and your body instinctively turns in the opposite direction to try and stay upright. This is the Optomotor Response (OMR).

For over a century, scientists thought this was a simple, robotic reflex: See spin → Turn opposite. Like a thermostat that turns on the AC the moment the room gets hot.

But this new study, looking at ants and earwigs, says: "Nope. It's much more complicated and interesting than that."

Here is the story of what they found, explained with some everyday analogies.

1. The "Ghost in the Machine" (Predictive Coding)

The researchers discovered that insects don't just react to what they see; they react to what they expected to see.

Think of your brain like a movie director.

• The Director (Your Brain): Sends a command to the camera crew (your eyes) saying, "I'm going to turn left."
• The Script (The Prediction): The director writes a script predicting exactly what the camera should see if the turn happens perfectly.
• The Reality (The Sensory Input): The camera actually records what is happening.

If the camera sees exactly what the script predicted, the director says, "All good, keep going." But if the camera sees something different—like the walls spinning faster than the script said—the director gets confused. That difference between the Script and the Reality is called a Prediction Error.

The study shows that insects don't just turn because they see spinning walls. They turn because their "director" realized, "Wait, I didn't write a script for the walls to spin this fast! Something is wrong!" They are trying to fix the error between their plan and reality.

2. The "Wobbly Compass" (Internal Oscillators)

You might think an insect's movement is smooth, like a car on a highway. But the study found that ants and earwigs actually wobble back and forth constantly, like a toddler learning to walk or a drunk sailor on a ship.

They have an internal "metronome" or oscillator in their brains that makes them naturally switch between turning left and right.

• The Old View: The spinning walls just push this wobbly compass in one direction.
• The New View: The spinning walls actually tweak the rhythm of the metronome. The insect isn't just being pushed; it's dancing to a beat that the spinning walls are changing.

3. The "Chaotic Dance" (Stochasticity)

Here is the wildest part. When the researchers put the insects in a situation where the visual world was spinning but the insect couldn't actually move (an "open-loop" setup), the insects didn't just turn steadily. They went crazy.

They would turn the "right" way for a bit, then suddenly spin the "wrong" way at high speed, then correct themselves. It looked chaotic.

The researchers realized this wasn't a glitch; it was stochastic (random).

• The Analogy: Imagine you are trying to walk through a crowded room. You have a plan to go straight. But because the room is crowded (noise), you occasionally bump into someone and stumble sideways.
• In the insect's brain, there is a little bit of "static" or noise. When the insect is confused by the spinning world, this noise causes it to occasionally take a wild guess and turn the opposite way. It's like the insect is saying, "I'm not sure what's happening, so I'll try turning the other way just to see what happens!"

4. The "Ancient Blueprint"

The most amazing part of this paper is that they tested two very different insects:

• Ants: Tiny, social, desert-dwelling navigators.
• Earwigs: Ancient, solitary, soil-dwelling creatures.

These two groups split from each other 350 million years ago. They are as different as a human is from a shark. Yet, they both use this exact same complex system of "Prediction Errors + Wobbly Metronomes + Random Guesses" to control their movement.

This suggests that this complex, predictive way of moving isn't a new invention; it's an ancient blueprint that has been in insect brains since the dawn of time.

The Big Takeaway

We used to think animals were like simple robots: Input (Spin) → Output (Turn).

This paper shows that animals are more like improvisational jazz musicians. They have a rhythm (the oscillator), they listen to the music (the visual world), they compare it to what they expected to hear (prediction), and if there's a mismatch, they throw in a random, wild note (stochasticity) to figure out what's going on.

The "Optomotor Response" isn't a simple reflex; it's the visible tip of a massive, complex, and ancient iceberg of brain power that allows insects to navigate a chaotic world.

Technical Summary

1. Problem Statement

The Optomotor Response (OMR) is a classic behavioral paradigm where animals turn in the direction of a rotating visual field to stabilize their gaze and prevent drifting. Traditionally, the OMR has been modeled as a deterministic, feedforward Stimulus-Response (S-R) mechanism (a reflex): the animal detects optic flow and generates a compensatory motor command.

However, this study challenges the S-R hypothesis by asking:

• Is the OMR a simple reflex, or does it emerge from a more complex predictive coding system involving efference copies (internal copies of motor commands used to predict sensory outcomes)?
• Is this mechanism conserved across phylogenetically distant insect lineages?
• How do internal neural oscillators and stochasticity interact with visual feedback to produce locomotion?

2. Methodology

The researchers employed a comparative approach using two distantly related insect species: ants (Cataglyphis velox) and earwigs (Forficula auricularia), representing over 350 million years of independent evolution.

Experimental Setup

• Virtual Reality (VR) Trackball: Insects were tethered to a floating polystyrene ball surrounded by a 360° cylindrical LED display. This allowed precise control of visual feedback while recording locomotion.
• Experimental Conditions:
  1. Open-Loop: The visual scene rotated at a constant speed (e.g., 180°/s for ants, 90°/s for earwigs) regardless of the insect's movement. The insect's motor commands did not update the visual scene.
  2. Closed-Loop: The insect's rotational movements on the trackball updated the visual scene (with a specific gain), simulating natural flight/walk conditions, while an additional constant external rotation was applied.
  3. Static/Dark: Control conditions with no rotation or no visual input.
• Data Acquisition: High-resolution tracking of angular velocity. Data was smoothed and filtered to isolate oscillatory patterns and "anti-optomotor" turns (turning against the rotation).

Analytical Approach

• Statistical Modeling: Linear Mixed-Effects Models (LMM) and Generalized Linear Mixed-Models (GLMM) were used to analyze angular velocity bias, oscillation frequency, and signal variation, treating individual identity as a random factor.
• Stochastic Analysis: The timing of "reversal events" (high-speed turns in the wrong direction) in open-loop conditions was analyzed to determine if they followed a predictable pattern or a geometric distribution (indicative of a random/stochastic process).
• Computational Modeling: A simple neural model was constructed to simulate the system. It included:
  • A sensory neuron detecting optic flow.
  • A prediction error neuron (difference between detected flow and predicted flow based on efference copy).
  • An intrinsic endogenous oscillator (mutually inhibiting left/right motor neurons) modulated by the prediction error.
  • Added neural noise to simulate stochasticity.

3. Key Results

A. OMR is Driven by Prediction Errors, Not Direct Optic Flow

• Hypothesis Test: If OMR were a simple S-R reflex based on detected optic flow, behavior should be stable in Open-Loop (where optic flow is constant) and variable in Closed-Loop (where optic flow varies with movement).
• Finding: The opposite occurred.
  • Closed-Loop: Insect behavior was stable and consistent.
  • Open-Loop: Behavior was highly variable and chaotic.
• Interpretation: In Open-Loop, the insect's motor commands generate a "predicted" optic flow that does not match the constant external rotation, creating a fluctuating prediction error. The insects' behavior tracks this error, not the constant visual stimulus. This confirms the OMR is driven by the computation of prediction errors.

B. Modulation of an Intrinsic Oscillator

• Both species displayed regular oscillations in angular velocity (alternating left/right turns) even during OMR.
• The frequency of these oscillations was modulated by the visual condition (static vs. rotating).
• This indicates the OMR is not a separate reflex pathway overriding the motor system, but rather a modulation of an endogenous neural oscillator by prediction error signals.

C. Stochasticity and "Anti-Optomotor" Turns

• In Open-Loop conditions, insects frequently turned at high speeds in the "wrong" direction (against the rotation).
• The intervals between these reversal events followed a geometric distribution, indicating they are stochastic (random) events rather than deterministic errors.
• The computational model successfully reproduced these complex dynamics, including the stochastic reversals, by adding noise to the intrinsic oscillator.

D. Conservation Across Lineages

• Despite vast evolutionary divergence, ants and earwigs exhibited identical behavioral patterns and statistical properties. This suggests that predictive coding combined with intrinsic oscillators is an ancestral feature of insect brains, conserved for >350 million years.

4. Key Contributions

1. Refutation of the Reflex Model: The study provides strong evidence that the OMR is not a simple, deterministic reflex but a complex, stochastic, closed-loop control process.
2. Identification of Mechanism: It demonstrates that the OMR emerges from the interaction between prediction errors (derived from efference copies) and endogenous neural oscillators.
3. Role of Stochasticity: It highlights that neural noise is not a bug but a feature, allowing insects to escape "attractors" (stable turning states) and explore different behavioral regimes.
4. Evolutionary Insight: It establishes predictive coding as a deeply conserved mechanism in the insect lineage, challenging the notion that such complex computation is unique to vertebrates.
5. Methodological Implication: It argues that neural recordings in fixed animals (open-loop) may capture prediction error signals rather than pure sensory detection, necessitating a re-evaluation of how neural data is interpreted.

5. Significance

This paper fundamentally shifts the understanding of insect visuo-motor control. It suggests that the brain does not merely react to the world but continuously predicts sensory consequences of its own actions. The "reflex" observed in averaged data is actually the byproduct of a dynamic, stochastic, closed-loop control system.

• Theoretical Impact: It bridges the gap between simple reflex models and complex predictive coding theories, showing they can coexist where the "reflex" is an emergent property of a predictive system.
• Neuroscience: It implies that neural feedback loops are ubiquitous in insect brains, and that "noise" plays a functional role in exploration and decision-making.
• Robotics/AI: The findings suggest that robust autonomous navigation systems should utilize predictive coding and intrinsic oscillators with stochastic elements, rather than relying solely on reactive feedback loops, to handle dynamic environments effectively.