The Experience Machine

1. Our brains are prediction machines, constantly anticipating sensory inputs

Brains are prediction machines that invoke external resources as easily (and for the same reasons) as they engage practical actions and activate different aspects of their own inner circuitry.

The predictive brain. Our brains are not passive receivers of information but active prediction engines. They constantly generate expectations about what we will see, hear, and feel based on past experiences and learned patterns. This predictive processing allows us to make sense of ambiguous or incomplete sensory data by filling in the gaps with our best guesses.

Efficiency through prediction. By anticipating sensory inputs, the brain can process information more efficiently. It only needs to focus on the differences between predictions and actual sensory data, rather than processing everything from scratch. This allows for rapid perception and decision-making in complex environments.

Examples of predictive processing:

• Hearing a familiar song on a bad radio and still understanding the lyrics
• Recognizing objects in partial view or poor lighting conditions
• Experiencing phantom phone vibrations due to frequent expectations

2. Experience arises from the interplay of predictions and sensory evidence

We see the world by predicting the world. But where prediction errors ensue, the brain must predict again.

Constructing reality. Our subjective experience of the world is not a direct reflection of reality, but a construction built from the interaction between our brain's predictions and incoming sensory evidence. When predictions match sensory input, we perceive the world as expected. When there's a mismatch, our brains update their models to better align with reality.

Precision-weighting. The brain assigns different levels of importance (precision) to predictions and sensory data based on their estimated reliability. This weighting determines how much influence each source has on our final perception. Attention can be understood as the process of adjusting these precision weights to focus on relevant information.

Implications:

• Optical illusions arise from the brain's predictions overriding ambiguous sensory data
• Our expectations can shape our perceptions, sometimes leading to biased interpretations
• Learning involves updating our predictive models to better match the world

3. Action is driven by self-fulfilling predictions of sensory consequences

To perceive is to find the predictions that best fit the sensory evidence. To act is to alter the world to bring it into line with some of those predictions.

Action as fulfilled prediction. In the predictive processing framework, actions are initiated by predicting their sensory consequences. The brain generates expectations of what it would feel like to perform an action, and the resulting prediction errors drive the motor system to make those predictions come true.

Unifying perception and action. This view provides a unified account of perception and action. Both involve minimizing prediction errors, either by updating internal models (perception) or by changing the world to match predictions (action). This explains the tight coupling between perception and action in skilled behaviors.

Key aspects of predictive action:

• Motor control as "controlled hallucination" of desired outcomes
• The role of proprioception (sense of body position) in action execution
• How practice improves performance by refining predictive models

4. Emotions and feelings emerge from bodily predictions and interoception

Everything that I see, hear, touch, and feel—so this new science suggests—reflects hidden wells of prediction.

Embodied emotions. Emotions are not just mental states but whole-body phenomena arising from the brain's predictions about internal bodily states (interoception). Our feelings of happiness, fear, or anger reflect the brain's best guess about our current physiological condition and its significance for our well-being.

Interoceptive predictive processing. The brain constantly generates predictions about internal bodily signals (heart rate, breathing, gut feelings) and compares them to actual sensory input. Mismatches between predicted and actual bodily states can drive emotional experiences and motivate actions to restore balance.

Implications for understanding emotions:

• The role of bodily feedback in shaping emotional experiences
• How expectations and context influence our interpretation of bodily sensations
• The potential for modulating emotions by altering interoceptive predictions

5. Psychiatric conditions can be understood as alterations in predictive processing

The upshot should be the start of a slow but important process eroding the old distinctions between psychiatry, neurology, and computational neuroscience, and at last embracing the fundamental unity of mind, body, and world.

A unified framework. Predictive processing offers a new way to understand various psychiatric and neurological conditions as disruptions in the brain's prediction machinery. This approach bridges the gap between mental and physical symptoms, providing a more integrated view of mental health.

Precision and psychopathology. Many disorders can be linked to imbalances in how the brain assigns precision to predictions versus sensory evidence. For example:

• Depression: Overly precise negative predictions about self and world
• Anxiety: Heightened precision of threat-related predictions
• Schizophrenia: Aberrant precision leading to hallucinations and delusions

Implications for treatment:

• Targeting predictive processes in therapy and medication
• Using cognitive interventions to reshape maladaptive predictions
• Exploring novel treatments that modulate precision-weighting in the brain

6. Our predictive minds extend beyond our skulls into the environment

Extended minds arise because predictive brains are naturally expert at exploiting opportunities to use information-gathering action loops to help them achieve their goals.

The porous mind. Our cognitive processes don't stop at the boundaries of our brains. The predictive mind readily incorporates external tools, technologies, and environmental scaffolding into its problem-solving strategies. This leads to an "extended mind" that spans brain, body, and world.

Cognitive offloading. We constantly use the environment to support our thinking and reduce cognitive load. Examples include:

• Using notebooks or smartphones as external memory
• Manipulating physical objects to aid problem-solving
• Leveraging social interactions for distributed cognition

Implications of extended cognition:

• Rethinking the nature of intelligence and cognitive boundaries
• Designing better tools and environments to augment human cognition
• Reconsidering how we understand and treat cognitive impairments

7. We can hack our predictive brains to improve our experiences and capabilities

Understanding the nature and effects of differing balances within the experience machine also locates neurotypical and atypical forms of human experience within a single, unifying framework, in ways that have significant implications for psychiatry, medicine, and clinical practice.

Leveraging prediction. By understanding how our brains construct experience through prediction, we can develop strategies to intentionally shape our perceptions, emotions, and behaviors. This opens up new possibilities for personal growth, therapy, and enhancing human potential.

Techniques for hacking the predictive brain:

• Placebo effects: Harnessing the power of expectations to improve health outcomes
• Reframing: Using language to alter the predictive context of experiences
• Meditation: Training attention to modulate precision-weighting of sensory inputs
• Virtual reality: Creating immersive environments to reshape predictive models
• Psychedelics: Temporarily disrupting entrenched predictions to enable new perspectives

Ethical considerations:

• Balancing the benefits of "hacking" with potential risks and unintended consequences
• Ensuring equitable access to cognitive enhancement technologies
• Respecting neurodiversity while addressing genuine impairments

Last updated: August 8, 2024