Your Brain Is (Almost) Perfect

1. Efficient computation drives biological systems

Recharge or die; this is the first law of efficient computation.

Energy efficiency shapes biology. Living organisms, from bacteria to humans, are driven by the need to compute efficiently. This principle arises from the fundamental constraint that all life runs on limited energy. Unlike modern computers that waste energy as heat, biological systems have evolved to minimize energy consumption while maximizing computational power.

Adaptations for efficiency:

• Slow and imprecise computations (compared to modern computers)
• Noisy neural signals that utilize available bandwidth
• Valuation mechanisms to prioritize energy allocation
• Compression of information and experiences

The brain's apparent limitations - slowness, imprecision, and noise - are actually sophisticated adaptations for energy efficiency. This efficiency allows complex cognition to run on just 20 watts of power, less than a dim light bulb.

2. The brain's imperfections are actually adaptations

The brain is a set of living computational devices that generate our mind, and it's almost miraculously efficient.

Evolved features, not bugs. What appear to be imperfections in brain function - slowness, imprecision, and noisiness - are actually highly optimized features. These characteristics allow the brain to operate with remarkable energy efficiency while still performing complex computations.

Adaptive brain features:

• Slow neural firing rates (20-60 Hz) compared to modern computers (billions of Hz)
• Imprecise computations that save energy by avoiding unnecessary precision
• Noisy neural signals that efficiently utilize available bandwidth
• Dynamic resource allocation (attention) to focus on the most important information

These features enable the brain to perform sophisticated computations while consuming minimal energy. This efficiency is crucial for survival, as the brain must operate continuously on limited resources.

3. Valuation mechanisms are fundamental to life

Nature has equipped biological computations with a measure of their value.

Valuing is caring. At the core of efficient biological computation is the ability to assign value to different options and outcomes. This valuation process allows organisms to make decisions that maximize survival and reproduction while minimizing energy expenditure.

Key aspects of biological valuation:

• Even single-celled organisms like E. coli can value and make decisions
• Valuation mechanisms guide energy allocation in the brain
• The ability to assign differential value to options enables choice
• Complex goal-seeking behavior emerges from these valuation systems

Valuation mechanisms are present at every level of biological systems, from molecules to social exchanges. This ability to care about outcomes and assign meaning to computations distinguishes biological computers from traditional machines.

4. Reward prediction errors guide learning and decision-making

Ideas act as reward signals from the point of view of the prediction error systems.

Learning from surprises. The brain's reward system, centered on dopamine neurons, operates on the principle of reward prediction errors. This system compares expected outcomes with actual outcomes, using the difference to guide learning and decision-making.

Key features of the reward prediction error system:

• Dopamine neurons encode the difference between expected and actual rewards
• This error signal drives learning and updates future predictions
• The system can assign value to abstract ideas and goals
• It enables both real and simulated (imagined) learning

This mechanism allows the brain to efficiently learn from experience and adapt to new situations. It also provides a framework for understanding how abstract ideas and cultural messages can gain motivational power by hijacking this system.

5. Ideas can hijack our reward systems

Sharks don't go on hunger strikes because they can't form and maintain ideas as we can.

The superpower of abstraction. Humans possess the unique ability to assign reward value to abstract ideas, allowing these concepts to guide behavior with the same force as primary rewards like food or sex. This capacity enables remarkable cognitive flexibility but can also lead to maladaptive behaviors.

Implications of idea-as-reward:

• Allows pursuit of long-term, abstract goals
• Enables cultural innovations and complex social structures
• Can lead to behaviors that override basic survival instincts (e.g., hunger strikes)
• Provides a framework for understanding ideological extremism and addiction

This mechanism explains how humans can pursue abstract goals, sometimes even at the cost of their own survival. It represents a powerful cognitive adaptation but also a potential vulnerability that can be exploited by cultural messages or lead to pathological behaviors.

6. Trust and regret are computational signals

You have to give someone the chance to cheat in order to learn to trust them.

Emotions as algorithms. Trust and regret, often thought of as purely emotional experiences, can be understood as computational signals that guide learning and decision-making in social contexts. These signals help individuals navigate complex social environments and learn from both real and imagined experiences.

Computational aspects of trust and regret:

• Trust: A signal that updates internal models of other individuals
• Regret: A learning signal based on the difference between actual and potential outcomes
• Both involve simulations of potential future scenarios
• These signals guide behavior in social exchanges and decision-making

Understanding trust and regret as computational processes provides insight into how the brain manages social relationships and learns from counterfactual experiences. This framework also offers new approaches to studying and potentially treating disorders involving social cognition.

7. Cultural messages exploit our valuation systems

We live in a sea of cultural messages, some more compelling than others.

The biology of marketing. Cultural messages, from advertising to ideologies, can gain power by exploiting the brain's valuation and reward prediction systems. These messages act as proxies for value, similar to how a light can come to predict juice in a classical conditioning experiment.

How cultural messages gain power:

• Associating with primary rewards or high-status indicators
• Exploiting the brain's tendency to assign value to predictive cues
• Utilizing the capacity for abstract ideas to act as rewards
• Tapping into social instincts and valuation mechanisms

This understanding provides insight into why certain cultural messages, brands, or ideologies can gain such strong influence over behavior. It also suggests potential vulnerabilities in our cognitive systems that can be exploited by malicious actors or lead to maladaptive behaviors.

8. Humans are not fully computable, but organized by computation

Evolution bumped into this same feature of our world and discovered a computation that specifies which collection of uncomputable properties to bring together to perform some function.

Computable organization of the uncomputable. While many aspects of human cognition and behavior can be described computationally, humans are not fully computable entities. Instead, evolution has produced a system that uses computable processes to organize and utilize uncomputable properties.

Implications of this view:

• Challenges simplistic computational theories of mind
• Suggests limits to artificial replication of human cognition
• Highlights the unique nature of biological computation
• Points to the need for new mathematical frameworks to understand cognition

This perspective reconciles the apparent computational nature of many cognitive processes with the difficulty of fully reducing human experience to computation. It suggests that understanding human cognition requires a more nuanced view that incorporates both computable and uncomputable aspects.

Last updated: November 8, 2024