The Real Reason for Brains: A Neuroscience of Movement
The Evolutionary Purpose of the Brain
Fundamental Question: Why do humans and other animals have brains?
Misconceptions: It is commonly but incorrectly assumed that brains evolved for perception or thinking.
The Movement Hypothesis: The brain exists for one reason and one reason only: to produce adaptable and complex movements. * Movement is the exclusive way humans have of affecting the world, with the only minor exception being sweating. * All forms of communication—speech, gestures, writing, and sign language—are mediated through the contraction of muscles.
Subservience of Other Processes: Sensory, memory, and cognitive processes are only important insofar as they drive or suppress future movements. * There is no evolutionary advantage to perceiving the color of a rose or storing childhood memories if they do not eventually impact motor behavior.
The Case of the Sea Squirt: * The sea squirt is a rudimentary animal with a nervous system that swims in the ocean during its juvenile life. * Once it implants on a rock (which it never leaves), the first thing it does is digest its own brain and nervous system for food. * Conclusion: Once the need for movement is gone, the brain becomes an unnecessary luxury.
The Difficulty of Movement Control: Humans vs. Machines
Current Progress: Neuroscience is doing poorly in understanding how the brain controls movement; it is a significantly harder problem than high-level cognition.
The Chess Analogy: * The Strategic Problem: Determining which piece to move where is a solved problem. IBM’s Deep Blue can defeat world champions like Garry Kasparov by looking at all possible moves and choosing the winning algorithm. * The Dexterity Problem: Picking up a chess piece and manipulating it is an unsolved problem. A five-year-old child easily outperforms the best robots in dexterity.
Robotics vs. Human Agility: * Manipulation Robotics: Current robotics is in the "dark ages." A PhD project from a top institute might train a robot to pour water into a glass, but it lacks human agility and cannot generalize to new tasks without years of further programming. * High Performance: Emily Fox holds a world record in cup stacking, involving the stacking and unstacking of cups against the clock. The internal brain processes allowing such speed and precision are currently unknown.
The Problem of Noise in Sensory and Motor Systems
Biological Signals: Controlling movement is difficult because biological signals are not clean; they are corrupted by "noise."
Definition of Noise: In engineering and neuroscience, noise refers to random variations that corrupt a signal (like the static on an old digital radio).
Types of Noise: * Sensory Noise: When placing a hand under a table, sensory feedback noise can cause a localization error of several centimeters. * Motor/Movement Noise: Motor output is variable; aiming for the same spot repeatedly (e.g., in darts) results in a spread of hits due to movement variability. * External Ambiguity: The world is variable; for example, a teapot might be full or empty, changing the requirements for movement.
Societal Value: Society places a high premium on individuals who can minimize the consequences of noise (e.g., professional golfers are rewarded with hundreds of millions of dollars for hitting a small ball into a hole).
Bayesian Decision Theory and Inference
Framework: The brain uses Bayesian decision theory to deal with uncertainty. It involves making inferences to generate "beliefs."
Representing Beliefs: Beliefs are represented by probabilities between (no belief) and (absolute certainty).
Sources of Information: 1. Data: Sensory input (vision, audition, etc.). 2. Prior Knowledge: Knowledge accumulated throughout life and stored in memory.
Bayesian Inference Formula: The brain uses mathematics to combine prior knowledge with sensory evidence to generate a new belief. * Tennis Example: When judging where a ball will land, a player uses sensory evidence (visual/auditory data showing a landing spot) and multiplies it by a "prior distribution" (knowledge of where an opponent typically hits the ball) to reach an optimal landing estimate.
Human Application: Research shows that humans are "Bayesian inference machines," learning the statistics of the world and the noise levels of their own senses to combine them optimally.
Prediction and the Neural Simulator
Anticipating the Future: To be Bayesian, the brain must predict the probability of different sensory feedbacks given its beliefs.
The Neural Simulator: The brain contains a neural simulator of the physical properties of the body and senses. * When a movement command is sent to the muscles, a copy of that command (an efference copy) is sent to the simulator to anticipate sensory consequences.
Distinguishing Internal vs. External Events: * Internal: Sensation caused by one's own movement (e.g., shaking a ketchup bottle). * External: Sensation caused by outside forces (e.g., someone tapping the ketchup bottle). * The brain subtracts the predicted internal sensation from the actual sensation; any discrepancy indicates an external event, which is more behaviorally relevant.
Evidence for Sensory Subtraction: Tickling and Force Escalation
The Tickling Experiment: * Hypothesis: You cannot tickle yourself because the brain predicts the sensation and subtracts it. * Method: A robotic interface where a subject's right-hand movement controls a robot that tickles their left palm. * Results: Introducing a time delay ( to seconds) between the movement and the tickle makes the sensation move from non-ticklish to equivalently ticklish as if another person (or robot) were doing it. * Conclusion: Sensory cancellation is tightly coupled with temporal causality.
The Force Escalation Experiment: * Scenario: Children in car backseats often get into fights that escalate in force because each child feels they were hit harder than they hit back. * Lab Test: Two players take turns applying the same force they just felt to each other's fingers via a force transfuser. * Initial Force: A motor applies an initial force of a quarter of a Newton (). * Observation: There is a escalation in force on every turn. * Mechanism: The person applying the force predicts and cancels the sensory consequence, underestimating the force they produced, while the recipient feels the full, uncancelled blow.
Movement Optimization and Path Selection
The Redundancy Problem: Tasks are symbolic (e.g., "I want to drink"), but the motor system must manage muscles and infinite possible paths and joint configurations.
Stereotypical Movement: Humans move in very specific, similar ways. The brain has dedicated neural circuitry to decode these patterns (biological motion). * People can identify age, gender, and emotion from just a few moving dots if they follow biological patterns.
Optimization Principle: Movements are planned to minimize the negative consequences of noise. * Noise/variability increases as force increases (Signal-dependent noise). * Humans choose paths that result in the smallest variability across many movements, effectively avoiding large forces to maintain precision.
Questions & Discussion
Chris Anderson: Does being a "movement chauvinist" mean that things like dreaming, yearning, and falling in love are just a side show or accident?
Daniel Wolpert: No, these processes are important because they eventually drive the right movement behaviors for survival and reproduction. Sensation and memory are only useful if they affect future movement; we likely forget most of childhood because it no longer influences our motor actions.
Insight into Consciousness: Studying vision or memory in isolation is a mistake; they must be studied with the realization that the movement system is the ultimate end-user of that information. For example, the brain uses vision very differently when it is intended to guide movement.