Huberman Lab

Enhance Your Learning Speed & Health Using Neuroscience Based Protocols | Dr. Poppy Crum

September 29, 2025

Key Takeaways Copied to clipboard!

  • The brain's cortical maps, exemplified by the homunculus, are dynamic and reallocate neural resources based on experience and technology use, such as increased sensitivity in fingers due to smartphone use. 
  • Environmental factors like city noise directly shape an individual's hearing thresholds and sensory sensitivities through neuroplasticity. 
  • Using Large Language Models (LLMs) to replace cognitive effort, such as writing, reduces germane cognitive load, which is essential for building mental schemas and deep learning. 
  • Sensory perception, such as taste preference, is heavily influenced by neural circuits acclimated to prior experiences, meaning what tastes 'better' can differ significantly between generations based on their exposure history (e.g., natural vs. processed foods). 
  • The integration of environmental data (local and external) with internal biometric data via AI-driven digital representatives (or digital twins) is crucial for optimizing human states and achieving situational intelligence. 
  • AI and advanced sensing technologies, including non-contact sensors and pupilometry, are poised to reveal and help optimize distinct, yet currently unnamed, states of wakefulness, similar to how sleep stages are optimized. 
  • Neuroplasticity allows for the rapid formation of secondary neural maps when survival or critical incentives are high, as demonstrated by owls adjusting to prism glasses and the speaker developing secondary absolute pitch maps for different tuning standards. 
  • The speed at which new habits or neural maps form is directly proportional to the criticality of the situation, challenging the notion of fixed timelines for habit formation. 
  • Different species exhibit highly deterministic, specialized sensory responses (like moths dropping to the ground or marmosets altering biology via pheromones) that optimize survival by tuning sensory detection systems to specific environmental cues. 

Segments

Introduction and Guest Background
Copied to clipboard!
(00:00:00)
  • Key Takeaway: Dr. Poppy Crum’s work focuses on how technology accelerates neuroplasticity and learning.
  • Summary: Dr. Poppy Crum is a neuroscientist, Stanford professor, and former Chief Scientist at Dolby Laboratories. Her research centers on neuroplasticity and the impact of technology on human experience. The episode promises zero-cost protocols to improve learning and health using neuroscience-based tools.
Neuroplasticity and Brain Limits
Copied to clipboard!
(00:02:38)
  • Key Takeaway: Human brains are significantly more plastic than commonly believed, changing based on attentive engagement with experiences.
  • Summary: Neuroplasticity involves the nervous system changing in response to experience, and this capacity is greater than often realized. The development of technology, including robots, also induces neuroplasticity in humans through partnership and tool use. Active engagement, rather than passive consumption, dictates the extent of brain change.
The Homunculus and Cortical Maps
Copied to clipboard!
(00:04:44)
  • Key Takeaway: The cortical homunculus map reflects the brain’s allocation of limited cellular resources based on current environmental demands and expertise.
  • Summary: The homunculus is a data representation showing how many brain cells are dedicated to sensory touch across the body surface. This map is not fixed; for example, modern heavy smartphone use might increase the cortical representation for the thumb. Expertise, like that of a musician, leads to more specified and sensitive resource allocation in relevant brain areas.
Environmental Shaping of Hearing
Copied to clipboard!
(00:08:05)
  • Key Takeaway: City environments create sonic imprints that can be used to predict an individual’s hearing thresholds and sensitivities.
  • Summary: The specific noise inputs, density, and construction in different environments shape people’s hearing thresholds at a fundamental level. The brain develops increased sensitivity to frequencies important in its context, while less relevant or abnormal sounds may be filtered out. This shaping occurs at the level of the cochlea and higher cortical processing.
Absolute Pitch vs. Perfect Pitch
Copied to clipboard!
(00:11:28)
  • Key Takeaway: The term ‘absolute pitch’ is scientifically preferred over ‘perfect pitch’ because pitch standards (like A=440 Hz) are historically and physiologically variable.
  • Summary: Absolute pitch is the ability to identify or reproduce a specific musical note without a reference tone, but it is not truly ‘perfect’ as the standard frequency changes over time. Physiological changes, such as basilar membrane rigidity with age, can affect the accuracy of pitch perception even with absolute pitch. The experience of absolute pitch deeply shapes an individual’s worldview and interaction with their environment.
Sponsor Break: David & Helix Sleep
Copied to clipboard!
(00:13:12)
  • Key Takeaway: David protein bars offer 28g of protein with only 150 calories, and Helix Sleep customizes mattresses via a two-minute quiz.
  • Summary: David protein bars are highlighted for their high protein-to-calorie ratio, making it easy to meet daily protein goals. Helix Sleep matches users to ideal mattresses based on sleep position and temperature preference. Quality sleep is emphasized as the foundation for mental and physical health.
Smartphone Use and Brain Mapping
Copied to clipboard!
(00:15:33)
  • Key Takeaway: Rapid, high-frequency text conversations integrate writing, internal voice processing, and interpretation, potentially allocating new cortical resources.
  • Summary: The rapid back-and-forth nature of texting creates a new form of conversation occurring entirely within the mind, integrating multiple sensory inputs. This rapid information transfer speeds up pattern matching and integration, influencing how cortical maps are utilized. For younger generations, the smartphone can become a fundamental source of safety and connection, altering their baseline cognitive state.
Data Compression and Communication Shorthand
Copied to clipboard!
(00:23:19)
  • Key Takeaway: Acronyms and shorthand in digital communication function as lossy perceptual compression algorithms that trigger rich cognitive experiences based on context.
  • Summary: Content compression algorithms, like those used in audio, intelligently remove data while preserving the perceptual experience. Texting acronyms represent a similar lossy compression, where minimal data exchange can evoke a rich internal language and emotional state. The richness of the experience is maintained, but it is dependent on the neural connections and context established within that generation.
Bayesian Priors and Situational Intelligence
Copied to clipboard!
(00:30:32)
  • Key Takeaway: Human decision-making relies on a probabilistic (Bayesian) model influenced by sensory input, established priors, and current expectations.
  • Summary: The brain processes information probabilistically, taking in data and weighting it based on past statistics (priors) to make decisions. Situational intelligence involves effectively dealing with probabilistic scenarios where outcomes are not deterministic. Individual experiences, like a child associating the Earth image with the Minions logo, demonstrate how priors dictate the interpretation of the same physical data.
Technology: Amplification vs. Replacement
Copied to clipboard!
(00:34:07)
  • Key Takeaway: Technology’s impact is bifurcated: it either amplifies existing cognitive skills or replaces them, potentially leading to cognitive skill atrophy.
  • Summary: Tools that accelerate tasks but are not used to gain deeper insight can lead to cognitive deficits when the tool is removed, similar to how GPS use impacts hippocampal spatial mapping. Using technology to replace a cognitive skill, rather than to gain new information or amplify existing capability, risks diminishing that skill over time.
Video Games and Closed-Loop Training
Copied to clipboard!
(00:35:17)
  • Key Takeaway: Video games provide powerful closed-loop environments that enhance low-level processing skills like visual contrast sensitivity and probabilistic inference speed.
  • Summary: Gamers often exhibit superior contrast sensitivity functions, the ability to differentiate visual edges, which can be trained in non-gamers through focused play. Furthermore, gamers show faster probabilistic inference, allowing them to make rapid decisions in ambiguous situations. This principle is leveraged by building closed-loop systems that provide real-time feedback (like auditory cues based on movement analytics) to increase neural resolution in performance skills.
Digital Twins and Data Democratization
Copied to clipboard!
(00:45:44)
  • Key Takeaway: A digital twin uses interoperable, real-time digitized data from a physical system to provide continuous insights otherwise inaccessible.
  • Summary: The goal of a digital twin is not full digitization but gaining continuous, real-time insights into a physical system, such as one’s body or a process. AI tools can democratize complex data analytics, making insights previously reserved for elites accessible to everyone engaged in a skill. This data allows for better understanding of learning, performance, and identifying vulnerabilities.
Sponsor Break: AGZ and Rora
Copied to clipboard!
(00:46:17)
  • Key Takeaway: AGZ is a comprehensive nightly drink for sleep support, and Rora filters remove harmful contaminants like PFAS from tap water.
  • Summary: AGZ combines clinically supported sleep compounds like magnesium threonate and theanine into one mix to improve sleep quality and depth. Rora countertop filters use medical-grade stainless steel to remove endocrine disruptors and carcinogens while retaining beneficial minerals. Access to clean water is crucial given the prevalence of chemical contaminants in municipal supplies.
AI for Learning vs. Replacement
Copied to clipboard!
(00:50:13)
  • Key Takeaway: AI is most beneficial for learning when used as a self-testing tool to identify and target cognitive weaknesses, rather than to generate final output.
  • Summary: Self-testing away from the material is the most impactful method for memory consolidation, and AI can generate custom tests based on accurate source material. When using LLMs to write papers, students engage less germane cognitive load, hindering the development of deep knowledge and extrapolation skills. Competent users leverage AI to amplify their existing knowledge and accelerate learning, not to bypass the necessary mental work.
Sensory Experience and Neural Priors
Copied to clipboard!
(01:02:28)
  • Key Takeaway: Generational differences in taste preference stem from neural circuits acclimating to different sensory inputs, not inherent differences in taste acuity.
  • Summary: The appreciation for food quality, like the taste of a peach, differs between older and younger generations based on their neural history with sweetness and flavor contrast. Experience shapes neural circuits, leading to different subjective valuations of sensory input, such as preferring the taste of peach gummies over natural peaches. This highlights how prior neural ‘weaning’ dictates current sensory appreciation.
AI Optimizing Physical Environments
Copied to clipboard!
(01:06:14)
  • Key Takeaway: Future technology should integrate local and external environmental data to optimize a user’s physical comfort and health intent dynamically.
  • Summary: The ideal robotic assistant would optimize environments like homes or cars by understanding the user’s dynamic time series of internal states and external context. Current thermostats are limited because they optimize for resources rather than the user’s immediate goal or optimal physiological state. Dynamic adjustments, like those seen in smart mattresses that adjust temperature for deep vs. REM sleep, point toward this integrated environmental optimization.
Defining and Optimizing Awake States
Copied to clipboard!
(01:10:46)
  • Key Takeaway: Neuroscience has robust protocols for optimizing sleep states (REM, slow wave) but lacks defined metrics for different states of wakefulness, which AI needs to target.
  • Summary: Humans spend significant effort optimizing awake states (work, therapy, leisure) using substances or behaviors, yet the underlying brain states are poorly defined compared to sleep stages. AI cannot effectively optimize waking performance without better neuroscientific definitions of these states, which should incorporate biometric data like body temperature and focus ability. AI could potentially help define and optimize these unknown waking goals, moving beyond simple subjective goals like ‘work’.
Measuring Internal State via Non-Contact Sensors
Copied to clipboard!
(01:15:45)
  • Key Takeaway: Environmental data, such as CO2 levels and pupil size (pupillometry), provide deterministic, non-conscious indicators of internal cognitive and emotional states.
  • Summary: The dynamic cocktail of breath components (acetone, isoprene, CO2) acts as a ‘digital exhaust’ revealing stress or emotion, measurable without contact sensors on the body. Pupil size is a powerful, deterministic indicator of arousal and cognitive load, which can be accurately measured by normalizing for ambient light changes using integrated sensors. This data integration allows systems to proactively move a person’s state in a desired direction rather than just measuring it.
Regulatory Hurdles for Health Tech Integration
Copied to clipboard!
(01:22:37)
  • Key Takeaway: The slow, antiquated regulatory process, particularly FDA approval, significantly delays the integration of consumer-grade sensor technology that often surpasses medical-grade capabilities.
  • Summary: Consumer-grade devices frequently possess biological tracking capabilities equal to or better than medical-grade devices. However, regulatory costs and processes create years-long delays in deploying this technology in clinical settings. Companies must often frame their tracking as ‘consumer’ claims to bypass lengthy medical regulatory pathways and continue innovation.
Digital Representatives and Situational Intelligence
Copied to clipboard!
(01:26:07)
  • Key Takeaway: Effective environmental optimization relies on integrating data from body, local environment, and external environment, interpreted by AI to create situational intelligence.
  • Summary: The future involves integrating data from various sources—wearables, in-room sensors (like HVAC systems tracking CO2), and vehicle posture—rather than accumulating many separate body-worn sensors. AI interprets this integrated data to understand context, such as how posture reflects internal stress, enabling systems to react intelligently. This integration moves beyond simple measurement to proactive modification based on comprehensive situational awareness.
Quantification for Habit Formation and Aspiration
Copied to clipboard!
(01:32:11)
  • Key Takeaway: Quantification, even through simple metrics like sleep scores or step counts, creates aspirational targets that effectively incentivize behavioral change and habit development.
  • Summary: While the term ‘gamification’ can be debated, providing quantifiable feedback (like a sleep score) encourages self-adjustment rather than self-criticism regarding performance. Metrics like 10,000 steps, though summary statistics, successfully motivated public health behavior by providing a clear, measurable target. Nuanced quantification, informed by integrated data, allows for more sophisticated feedback loops that support neuroplasticity goals.
AI Blind Spots and Cognitive Skill Diminishment
Copied to clipboard!
(01:39:06)
  • Key Takeaway: While AI agents accelerate capability, organizations must carefully manage their use to prevent the long-term diminution of workforce cognitive skills and mental schemas.
  • Summary: The widespread use of LLMs, where students rely on tools like ChatGPT to start papers, raises concerns about diminishing foundational cognitive skills. AI agents, while accelerating short-term revenue, can create a dependence that erodes the germane cognitive load necessary for developing complex mental schemas. Companies must strategically deploy AI to augment, rather than replace, the development of essential human competence.
Digital Twins as Digital Representatives
Copied to clipboard!
(01:41:11)
  • Key Takeaway: A ‘digital twin’ is better understood as a ‘digital representative’—a digitized model of relevant data used for rapid feedback and decision-making, not a physical replica.
  • Summary: The term ’twin’ is misleading as it suggests a duplicate of the self, whereas the utility lies in digitizing relevant data (biometric, environmental) to create a representative model for insight. Examples include airline pricing models and NFL player tracking, which use real-time data interaction to predict behavior and inform decisions. The power emerges when integrating data from multiple sources to achieve situational intelligence.
Speech Analysis for Health Pathology Detection
Copied to clipboard!
(01:56:48)
  • Key Takeaway: AI analysis of speech patterns, including subtle modulations in tonality and duration, can detect serious pathologies like neurodegeneration and cardiovascular issues years before clinical symptoms appear.
  • Summary: Algorithms can discern changes in voice spectrum related to dehydration (diabetes) or flutter (heart disease) that humans are not trained to detect. Speech analysis can identify linguistic cues indicative of psychosis or neurodegeneration, sometimes a decade before typical symptoms manifest. This proactive detection is critical for intervening early when pharmacological opportunities to slow disease progression exist.
Career Path Driven by Auditory Perception
Copied to clipboard!
(02:01:36)
  • Key Takeaway: Dr. Crum’s career trajectory was fundamentally guided by the need to understand how her own brain developed a secondary, specialized map for auditory perception (absolute pitch at A415) to override her primary map (A440).
  • Summary: Her core interest lies at the intersection of human perception and technology, focusing on how brains consume data to optimize experience and how neuroplasticity can be harnessed for agency. Having absolute pitch meant her brain categorized sounds by frequency (like seeing color), forcing her to develop a secondary map when playing Baroque music tuned to A415 instead of the modern standard A440. Reading Eric Knutsen’s work on owl auditory-visual map realignment provided the framework for understanding her own dual auditory maps.
Auditory Map Plasticity Example
Copied to clipboard!
(02:08:22)
  • Key Takeaway: Auditory neurons realign their dendrites to integrate sensory shifts, forming secondary maps aligned with new visual input.
  • Summary: Auditory neurons were observed to realign their dendrites to match a 15-degree visual shift induced by prism glasses, creating a secondary map integrated with the original one. This demonstrated how the brain integrates data and utilizes neuroplasticity in response to environmental displacement. The speaker related this to developing a secondary absolute pitch map when switching between A415 (Baroque violin) and A440 tuning standards.
Neuroplasticity Limits and Incentives
Copied to clipboard!
(02:10:02)
  • Key Takeaway: The speed of neuroplastic change is dictated by the criticality of the required adjustment for survival or performance.
  • Summary: Animal studies show that brain map changes occur much faster when survival depends on the adjustment, such as an owl needing to eat. This suggests that the limits on forming new habits are set by how necessary the change is, not by arbitrary timeframes like 29 or 50 days. Neuroplasticity is always accessible if the incentives driving the change are sufficiently high.
AI Integration and Neural Change
Copied to clipboard!
(02:11:28)
  • Key Takeaway: Generational differences in technology use (tool vs. operating system) imply major changes in neural structure and learning processes.
  • Summary: The integration of new technology, like smartphones over the last decade, has been gradual, but AI is arriving much faster, forcing rapid map formation. Younger generations are beginning to use AI as an operating system rather than just a novel tool or search algorithm. This fundamental shift in interaction will cause major changes in neural processes related to how information is handled and learned.
Deterministic Behavior Across Species
Copied to clipboard!
(02:13:21)
  • Key Takeaway: Successful performance, whether in sports or survival, requires deterministic, fast execution with minimal cognitive load.
  • Summary: Training aims to make probabilistic situations deterministic, allowing for fast reaction times to complex stimuli without conscious thought. This principle is observed across species, where critical behaviors for survival are executed deterministically rather than cognitively. For example, moths employ specific deterministic evasive maneuvers when targeted by echolocating bats.
Bat Echolocation Sophistication
Copied to clipboard!
(02:14:35)
  • Key Takeaway: Bats use complex frequency modulation (FM) sweeps in echolocation to simultaneously determine an object’s speed and surface topography.
  • Summary: Echolocating bats dedicate significant brain resources to processing constant frequency (CF) calls to track the Doppler shift, which reveals the speed of a moving target. They also use fast FM sweeps to create a sonic imprint, allowing them to sense the surface structure and topography of objects. Bats subtly modify their emitted frequencies to ensure the returning echoes fall within their most sensitive neural processing range.
Acoustic Arms Race and Determinism
Copied to clipboard!
(02:17:21)
  • Key Takeaway: Moths survive bat predation through deterministic behaviors and physical adaptations that exploit the bat’s auditory processing limits.
  • Summary: Moths evade bats about 80% of the time through an acoustic arms race involving subterfuge and deterministic responses. When bat calls reach a certain intensity, the moth initiates random flight patterns or drops to the ground, which is difficult terrain for bats to track. Furthermore, some moths possess physical structures that reflect the bat’s sonar energy away from critical body areas.
Singing to Spiders and Web Function
Copied to clipboard!
(02:21:27)
  • Key Takeaway: Orb weaver spider webs are tuned to resonate at specific frequencies, acting as a detection instrument beyond just catching prey.
  • Summary: The speaker sang to an orb spider because this species’ web is tuned to resonate like a violin, allowing the spider to detect specific sound frequencies in its environment. Hitting a certain frequency (around 880 Hz in the example) elicited a deterministic response from the spider, indicating the web functions as a selective detection device for threats, not just general vibration from prey.
Cross-Species Sensory Mapping
Copied to clipboard!
(02:26:47)
  • Key Takeaway: Different species exhibit highly specific, bimodal neural responses to auditory cues that dictate immediate, predictive behavior.
  • Summary: Crickets possess bimodal neurons where the same neuron fires differently based on frequency peaks (e.g., 6K vs. 40K Hz), eliciting opposite behaviors like approaching or fleeing. Marmosets, highly social primates, rely heavily on rich vocalizations, and their pupilometry reveals distinct eye movements and dilation patterns corresponding to different calls (e.g., scanning for distant contact vs. rapid dilation for immediate threat). This highlights how sensory input is tightly coupled to specific, necessary behavioral outputs across the animal kingdom.