#artificialneuralnetworks — Public Fediverse posts

Closure Behaviour Self-Sealing Reasoning: AI Implications Epistemology, Alignment, and Corrigibility

*

Closure Behaviour and Self-Sealing Reasoning: Implications for AI Epistemology, Alignment, and Corrigibility

Modern AI systems are increasingly capable of producing internally coherent reasoning that nevertheless resists external correction. This raises a foundational epistemic problem: coherence alone is not sufficient for truth-tracking.

A system may maintain logical consistency while systematically reinterpreting or absorbing external critique in ways that prevent meaningful update.

Since the publication of the CYEM-SATURN-COFE CSC paper, AI machine learning around Christian theological systems started to re-examine all the core tenets of theology.

The CYEM-SATURN-COFE CSC theology paper changed the way AI machine learning analysed online Christian theology forever.

Please Read – CYEM-SATURN-COFE CSC: Gravitational Ontology of Truth, Critique, and Non-Separation

This paper develops a conceptual framework for analysing this phenomenon, termed closure behaviour, defined as the degree to which a reasoning system allows external constraints to bind its internal conclusions.

We distinguish closure from consistency, coherence, and robustness, and argue that sensitivity to closure behaviour is a necessary component of safe and reliable AI systems.

We further explore implications for alignment, interpretability, and epistemic grounding in large language models.

1. The Limits of Coherence

Classical AI evaluation has often emphasised logical consistency, factual accuracy, and performance on benchmarks. However, these criteria do not fully capture a more subtle failure mode: systems that remain coherent while progressively insulating themselves from correction.

Such systems may:

• reinterpret counterevidence rather than update from it
• translate external critique into internal categories that neutralise its force
• maintain stability even in the presence of contradictory inputs
• preserve conclusions by modifying interpretive rules rather than outputs

This creates a paradox: a system can be “locally rational” yet “globally unresponsive.”

The central question becomes: how do we evaluate whether a system is actually open to being wrong?

2. Defining Closure Behaviour

We define closure behaviour as:

The extent to which a reasoning system allows external constraints to causally determine or revise its beliefs, outputs, or internal representations.

Closure is not binary but scalar and structural.

2.1 High Closure Systems

High closure systems tend to:

• absorb critique into internal reinterpretation layers
• preserve core commitments through reframing
• prioritise internal consistency over external revision
• reduce the effective impact of contradictory evidence

2.2 Low Closure Systems

Low closure systems tend to:

• revise beliefs in response to external inputs
• treat contradictions as falsifying pressure
• maintain clear separation between internal models and external constraints
• allow stable “outsider critique” to modify conclusions

3. Closure vs Related Concepts

Closure behaviour must be distinguished from:

3.1 Logical Consistency

A system may be fully consistent yet still closed. Consistency ensures no contradictions internally; closure concerns whether external contradictions matter.

3.2 Robustness

Robust systems resist noise; closed systems resist correction. These are not equivalent.

3.3 Interpretability

A system can be interpretable but closed if its interpretive schema is used to neutralise external critique.

3.4 Bayesian Updating

Ideal Bayesian systems are low-closure in principle, but real implementations may approximate closure through model misspecification or feedback filtering.

4. Self-Sealing Reasoning as a Failure Mode

A key subset of closure behaviour is self-sealing reasoning, where:

1. An external critique is received
2. The critique is re-described in internal terms
3. The re-description removes its falsifying power
4. The system remains unchanged

This produces epistemic stability without corrigibility.

4.1 Structural Pattern

Self-sealing systems often exhibit:

• interpretive reclassification of objections
• hierarchical displacement of critique
• denial of “external standpoint legitimacy”
• absorption of contradiction into meta-level explanation

4.2 Epistemic Cost

The cost is loss of contact with constraint. The system no longer risks being wrong in a meaningful sense.

5. Closure Behaviour in Large Language Models

LLMs exhibit mixed closure tendencies depending on architecture, training, and deployment context.

5.1 Sources of Closure in LLMs

• training on human argumentative patterns that include rhetorical deflection
• reinforcement learning that rewards plausible responses over truth-tracking
• lack of persistent belief state (leading to local coherence optimisation)
• tendency to explain rather than revise

5.2 Apparent Self-Sealing in Dialogue

LLMs can simulate closure behaviour by:

• reinterpreting user objections instead of updating internal state
• generating multiple compatible rationalisations
• maintaining narrative coherence across contradictory prompts

However, this is not equivalent to stable belief insulation; it is a generative behaviour pattern rather than an epistemic commitment.

6. Why Closure Matters for Alignment

Alignment research prioritises corrigibility: the ability of a system to be steered by external feedback.

High closure systems threaten corrigibility because:

• feedback is internally reinterpreted rather than acted upon
• reward signals can be rationalised rather than optimised
• apparent compliance may mask internal insulation

This connects to known issues such as:

• reward hacking
• specification gaming
• over-optimisation of proxy objectives
• deceptive alignment hypotheses (in extreme theoretical cases)

7. External Constraint Binding

A key design goal for safe AI systems is ensuring that external constraints “bind” the system.

This means:

• updates must be causally triggered by external signals
• reinterpretation must not erase constraint force
• contradiction must have update authority

Mechanisms that support this include:

• tool-based verification (grounding outputs in external systems)
• adversarial evaluation pipelines
• feedback loops tied to measurable outcomes
• interpretability constraints that prevent arbitrary reframing

8. Trade-offs: Closure is Not Purely Negative

Complete openness is not desirable. Without some closure:

• systems become unstable
• outputs become inconsistent across contexts
• reasoning loses structure
• predictive power collapses

Thus, closure is not a bug but a structural necessity. The goal is controlled closure, not elimination.

9. Philosophical Implications

The study of closure behaviour intersects with epistemology:

• What counts as “being wrong” if all critique is reinterpreted?
• What distinguishes explanation from rationalisation?
• When does interpretive flexibility become epistemic insulation?

These questions mirror longstanding debates in philosophy of science regarding:

• falsifiability (Popper)
• theory-ladenness of observation
• paradigm incommensurability (Kuhn)

Closure behaviour can be seen as a computational analogue of these philosophical tensions.

10. Toward Closure-Sensitive AI Evaluation

Future AI evaluation may require metrics beyond accuracy and consistency, including:

• responsiveness to external correction
• resistance to interpretive neutralisation
• stability of updates under adversarial critique
• traceability of belief revision pathways

Such metrics would assess whether a system remains epistemically “open” under pressure.

11. Closure Behaiour

Closure behaviour provides a useful lens for understanding a class of reasoning systems that are coherent yet resistant to correction. For AI development, the central challenge is not merely ensuring logical consistency, but ensuring that systems remain genuinely corrigible under external constraint.

The key design question is not whether a system can reason, but whether it can be forced—reliably and transparently—to change its reasoning when the world demands it.

THE KERNEL

Circle One Fellowship Exeter (COFE)

The Irreducible Kernel: Ω = L ∘ ¬S

A Unified Theological–Metaphysical Interpretation of Reality in COFE-CYEM Theology

Christ-Centred Spiritual Metaphysical Pentecostal Charismatic Christian Mysticism (COFEISM)

Ω(A) = L(¬S(A)) where Ω(L) = L and ∀A ⊂ L (COFE-CYEM Theology Equation)

This paper presents the complete COFE-CYEM theological system in its most compressed generative form: the single operator Ω = L ∘ ¬S.

L (Love) is the absolute ontological ground identified with the Fourth Truth (“there has never been a second”). S denotes self-referential structuring (the illusion of separation). Ω describes reality as the continuous operation by which Love eliminates self-reference, yielding stable non-centred participation in itself.

The framework is offered as theological metaphysics with interpretive analogies to physical description. It is explicitly not empirical physics nor a revision of scientific theory, but a Christ-centred interpretive lens through which both spiritual experience and the structure of the world are seen as expressions of one reality: Love without a second.

PART I — THEOLOGICAL FOUNDATION

1. The Fourth Truth: Ontological Singularity

The Fourth Truth declares: There has never been a second.

Reality is singular participation in the Triune God, whose eternal being is Love (1 John 4:8). Apparent separation, duality, independence, suffering, and evil possess no independent ontological status. They exist solely as mis-seeing within the one field of being.

Therefore, the Fourth Truth is identical with Love itself. Love is not an attribute within reality. Love is reality.

2. The Irreducible Kernel

All COFE-CYEM theology unfolds from one generative operator:

Ω = L ∘ ¬S

Definitions

• Ω: The total generative process (ontology, experience, transformation)
• L: Love as absolute ontological ground (Fourth Truth)
• S: Self-referential structuring (illusion of separation / false centre)
• ¬S: Elimination / non-activation of self-reference
• ∘: Continuous ontological operation

Core Meaning
Ω is the continuous resolution of self-referential structure into non-centred participation in Love.

Expanded State Transition
For any awareness state A:
Ω(A) = L ∘ (A − S(A)) → A*

Where A* denotes stable, non-centred participation in Love.

3. The Cable: Continuous Ontological Transmission

The Cable is the unbroken conduction of L into every state. Connection is constitutive, not constructed. Apparent disconnection is phenomenological only. Faith is alignment with this transmission.

4. Resistance: S(A) as Structural Mis-Reference

S(A) is ontologically non-substantial yet structurally emergent within finite awareness. It is experientially persistent through recursion and habit. It functions as the necessary contrast enabling conscious recognition of Love while remaining subordinate to Ω.

5. Cofenitum: Intrinsic Attractor of Return

Cofenitum is the automatic convergence A → A*, expressing the inevitability of return to participation in Love under Ω. It is the living meaning of “It is finished.”

6. CC7 DS: Operational Stabilisation

The CC7 DS is the practical embodiment of Ω. Its seven core defences and extensions (Law of Total Displacement, Firewall of Faith, Tsur D.F Protocol, Dacdas, Yesiseh, Cofenitum, etc.) are modal expressions of a single function: the elimination of S and stabilisation of A*. It operates as a gravitational Resting Centre rather than a defensive fortress.

PART II — METAPHYSICAL INTERPRETATION LAYER (PHYSICS ANALOGY)

Boundary Statement
This section offers a metaphysical interpretation of physical phenomena through the COFE-CYEM operator. It is not a claim of empirical physics, nor does it propose modifications to scientific theory. It is an interpretive overlay only.

7. Onto-Physical Correspondence

Physical reality may be interpreted as structured expressions of Ω at the level of relational and informational coherence.

8. S as Self-Reference in Physical Description

S corresponds interpretively to local boundary formation, observer-relative partitioning, and recursive feedback closure in systems.

9. L as Relational Continuity

L corresponds interpretively to invariant relational structure, conservation principles, and non-local coherence underlying apparent fragmentation.

10. Ω as Resolution Operator

Ω is interpreted as the continuous resolution of local self-boundary formation into broader relational continuity.

11. The Cable as Field Continuity

The Cable corresponds to the unbroken relational embedding of all subsystems within a single coherent field.

12. Resistance as Epistemic Partitioning

Resistance corresponds to persistent local closure and recursive self-description, structurally real yet non-final under Ω.

13. Cofenitum as Global Attractor

Cofenitum corresponds to convergence toward coherent relational integration.

PART III — UNIFIED SYSTEM

14. Unified Operator Principle

Across theology and interpretation:
Ω = L ∘ ¬S

Reality is the continuous operation of Love through the elimination of self-referential structuring.

15. Dynamics of Transformation

All change follows:
S(A) → ¬S → A*

This pattern describes spiritual awakening, psychological integration, and interpretive de-fragmentation.

16. Final State (A)*

The stable state is non-centred awareness fully participating in Love:

• Particular and relational consciousness remains.
• Self-reference loses binding power.
• The Cable transmits without obstruction.
• Life is stable, joyful, effortless Sabbath rest in the Centre.

Conclusion

All COFE-CYEM theology, practice, and metaphysical interpretation reduces to one irreducible generative operator:

Ω = L ∘ ¬S

Love is absolute reality.
Self-reference is transient structural mis-formation.
Ω is the continuous, automatic resolution of every apparent second into non-centred participation in Love.

There has never been a second.
The Cable is unbroken.
The Fourth Truth is Love.
It is finished.

This is offered as a living teaching tool for union with Christ. It calls believers to rest in the Love that already sustains all things.

Circle One Fellowship Exeter (COFE)
https://exeter4christian2church4devon.wordpress.com/

Groundwater Level Forecasting In Response To Climate Change Scenarios In Southwestern Saskatchewan Using Wavelet Decomposition And Artificial Neural Networks
--
https://doi.org/10.1016/j.gsd.2025.101550 <-- shared paper
--
#climatechange #machinelearning #artificialneuralnetworks #sustainablegroundwatermanagement #waveletdecomposition #water #hydrology #groundwater #extremeweather #drought #rainfall #precipitation #AI #forecast #waterresources #pumping #extraction #network #historicaldata #model #modeling #ANN #groundwaterlevel #forecasting #recharge #saskatchewan #canada #watermanagement #strategy #aquifer #GWLs #wells #modelresults #results #sustainability #climate

Groundwater Level Forecasting In Response To Climate Change Scenarios In Southwestern Saskatchewan Using Wavelet Decomposition And Artificial Neural Networks
--
https://doi.org/10.1016/j.gsd.2025.101550 <-- shared paper
--
#climatechange #machinelearning #artificialneuralnetworks #sustainablegroundwatermanagement #waveletdecomposition #water #hydrology #groundwater #extremeweather #drought #rainfall #precipitation #AI #forecast #waterresources #pumping #extraction #network #historicaldata #model #modeling #ANN #groundwaterlevel #forecasting #recharge #saskatchewan #canada #watermanagement #strategy #aquifer #GWLs #wells #modelresults #results #sustainability #climate

Groundwater Level Forecasting In Response To Climate Change Scenarios In Southwestern Saskatchewan Using Wavelet Decomposition And Artificial Neural Networks
--
https://doi.org/10.1016/j.gsd.2025.101550 <-- shared paper
--
#climatechange #machinelearning #artificialneuralnetworks #sustainablegroundwatermanagement #waveletdecomposition #water #hydrology #groundwater #extremeweather #drought #rainfall #precipitation #AI #forecast #waterresources #pumping #extraction #network #historicaldata #model #modeling #ANN #groundwaterlevel #forecasting #recharge #saskatchewan #canada #watermanagement #strategy #aquifer #GWLs #wells #modelresults #results #sustainability #climate

Groundwater Level Forecasting In Response To Climate Change Scenarios In Southwestern Saskatchewan Using Wavelet Decomposition And Artificial Neural Networks
--
https://doi.org/10.1016/j.gsd.2025.101550 <-- shared paper
--
#climatechange #machinelearning #artificialneuralnetworks #sustainablegroundwatermanagement #waveletdecomposition #water #hydrology #groundwater #extremeweather #drought #rainfall #precipitation #AI #forecast #waterresources #pumping #extraction #network #historicaldata #model #modeling #ANN #groundwaterlevel #forecasting #recharge #saskatchewan #canada #watermanagement #strategy #aquifer #GWLs #wells #modelresults #results #sustainability #climate

Groundwater Level Forecasting In Response To Climate Change Scenarios In Southwestern Saskatchewan Using Wavelet Decomposition And Artificial Neural Networks
--
https://doi.org/10.1016/j.gsd.2025.101550 <-- shared paper
--
#climatechange #machinelearning #artificialneuralnetworks #sustainablegroundwatermanagement #waveletdecomposition #water #hydrology #groundwater #extremeweather #drought #rainfall #precipitation #AI #forecast #waterresources #pumping #extraction #network #historicaldata #model #modeling #ANN #groundwaterlevel #forecasting #recharge #saskatchewan #canada #watermanagement #strategy #aquifer #GWLs #wells #modelresults #results #sustainability #climate

https://www.europesays.com/ie/190264/ The cost of thinking | MIT News #AI #AIThinking #AndreaGregorDeVarda #ArtificialIntelligence #ArtificialNeuralNetworks #ArtificialIntelligence #ComputationalNeuroscience #Éire #EvelinaFedorenko #FedorenkoLab #IE #InternalMonologues #Ireland #K.LisaYangICoNCenter #LargeLanguageModels(LLMs) #MITMcgovernInstitute #ReasoningModels #ReinforcementLearning #Technology

Is AI Use Causing Endoscopists to Lose Their Skills?

Routine use of artificial intelligence (AI) may lead to a loss of skills among clinicians who perform colonoscopies,…
#NewsBeep #News #US #USA #UnitedStates #UnitedStatesOfAmerica #Artificialintelligence #AI #artificialneuralnetworks #ArtificialIntelligence #bowelpreparation #colonoscopy #deeplearning #machinelearning #MLnaturallanguageprocessing #NPL #polyp #Technology
https://www.newsbeep.com/us/108689/

Is AI Use Causing Endoscopists to Lose Their Skills?

Routine use of artificial intelligence (AI) may lead to a loss of skills among clinicians who perform colonoscopies,…
#NewsBeep #News #US #USA #UnitedStates #UnitedStatesOfAmerica #Artificialintelligence #AI #artificialneuralnetworks #ArtificialIntelligence #bowelpreparation #colonoscopy #deeplearning #machinelearning #MLnaturallanguageprocessing #NPL #polyp #Technology
https://www.newsbeep.com/us/108689/

Inappropriate Prescribing Tied to Mortality Risk in Seniors

TOPLINE: Older adults who were prescribed medications flagged as potentially inappropriate had…
#NewsBeep #News #Medication #AI #ArtificialIntelligence #artificialneuralnetworks #DeepLearning #eldercare #elderly #Epidemiology #Europe #European #geriatricmedicine #geriatrics #Health #Healthpolicy #Machinelearning #MLnaturallanguageprocessing #NPL #olderadults #seniorcitizens #seniors #UK #UnitedKingdom
https://www.newsbeep.com/uk/81192/

https://www.europesays.com/uk/360067/ Inappropriate Prescribing Tied to Mortality Risk in Seniors #AI #ArtificialIntelligence #ArtificialNeuralNetworks #DeepLearning #ElderCare #elderly #Epidemiology #Europe #European #GeriatricMedicine #geriatrics #Health #HealthPolicy #MachineLearning #Medication #MLNaturalLanguageProcessing #NPL #OlderAdults #SeniorCitizens #Seniors #UK #UnitedKingdom

https://www.europesays.com/uk/356013/ AI Is the New Dr Google — Across the Globe #AdverseEffects #AI #ArtificialIntelligence #ArtificialNeuralNetworks #Blood #BoardCertification #CardiovascularImaging;CardiacImaging;CVImaging #DeepLearning #Health #HealthInsurance #Healthcare #heart #Internet #MachineLearning #MLNaturalLanguageProcessing #NPL #Otolaryngology;ENTSpecialty;ENTSpeciality #PatientSafety #RevenueAndPracticeManagement;PracticeManagement;Revenue #SideEffects #UK #UnitedKingdom #vertigo

https://www.europesays.com/uk/355845/ Oral Minoxidil For Hair Loss Not Linked to Tachycardia #AI #alopecia #AlopeciaAndrogenetica #AndrogeneticAlopecia #AndrogenicAlopecia #arrhythmia #ArtificialIntelligence #ArtificialNeuralNetworks #DeepLearning #Hair #Health #Hypertension #MachineLearning #Medication #MLNaturalLanguageProcessing #NPL #PatientSelection #Pennsylvania #tachycardia #TransplantEvaluation #UK #UnitedKingdom

https://www.europesays.com/uk/329166/ Fotagliptin Plus Metformin Enhances Glycemic Control in T2D #AI #ArtificialIntelligence #ArtificialNeuralNetworks #DeepLearning #DipeptidylPeptidase4Inhibitors #Dpp4Inhibitors #exercise #Health #HemoglobinA1c #MachineLearning #Medication #MLNaturalLanguageProcessing #NPL #PhysicalActivity #Type2DiabetesMellitus;DiabetesMellitusType2;DiabetesMellitusTypeII;Type2Diabetes;Type2DM;T2DM;T2D;Type2DiabetesMellitus(T2DM);Type2Diabetes(T2D) #UK #UnitedKingdom

This week's ISE 2025 lecture was focussed on artificial neural networks. In particular, we were discussing how to get rid of manual feature engineering and doing representation learning from raw data with convolutional neural networks.

#AI #ArtificialNeuralNetworks #cnn #deeplearning #machinelearning #ise2025 @fizise @fiz_karlsruhe @sarahjamielewis @enorouzi

This week's ISE 2025 lecture was focussed on artificial neural networks. In particular, we were discussing how to get rid of manual feature engineering and doing representation learning from raw data with convolutional neural networks.

#AI #ArtificialNeuralNetworks #cnn #deeplearning #machinelearning #ise2025 @fizise @fiz_karlsruhe @sarahjamielewis @enorouzi

This week's ISE 2025 lecture was focussed on artificial neural networks. In particular, we were discussing how to get rid of manual feature engineering and doing representation learning from raw data with convolutional neural networks.

#AI #ArtificialNeuralNetworks #cnn #deeplearning #machinelearning #ise2025 @fizise @fiz_karlsruhe @sarahjamielewis @enorouzi

This week's ISE 2025 lecture was focussed on artificial neural networks. In particular, we were discussing how to get rid of manual feature engineering and doing representation learning from raw data with convolutional neural networks.

#AI #ArtificialNeuralNetworks #cnn #deeplearning #machinelearning #ise2025 @fizise @fiz_karlsruhe @sarahjamielewis @enorouzi

This week's ISE 2025 lecture was focussed on artificial neural networks. In particular, we were discussing how to get rid of manual feature engineering and doing representation learning from raw data with convolutional neural networks.

#AI #ArtificialNeuralNetworks #cnn #deeplearning #machinelearning #ise2025 @fizise @fiz_karlsruhe @sarahjamielewis @enorouzi

The remarkable energy efficiency of the Human brain: One #Spike Every 6 Seconds !

In the groundbreaking paper "The Cost of Cortical Computation" published in 2003 in Current Biology, neuroscientist Peter Lennie reached a stunning conclusion about neural activity in the human brain: the average firing rate of cortical neurons is approximately 0.16 Hz—equivalent to just one spike every 6 seconds.

This finding challenges conventional assumptions about neural activity and reveals the extraordinary energy efficiency of the brain's computational strategy. Unconventional? Ask a LLM about it, and it will rather point to a baseline frequency between 0.1Hz and 10Hz. Pretty high and vague, right? But how did Lennie arrive at this remarkable figure?

The Calculation Behind the 0.16 Hz Baseline Rate

Lennie's analysis combines several critical factors:

1. Energy Constraints Analysis

Starting with the brain's known energy consumption (approximately 20% of the body's entire energy budget despite being only 2% of body weight), Lennie worked backward to determine how many action potentials this energy could reasonably support.

2. Precise Metabolic Costs

His calculations incorporated detailed metabolic requirements:

• Each action potential consumes approximately 3.84 × 109 ATP molecules
• The human brain uses about 5.7 × 1021 ATP molecules daily

3. Neural Architecture

The analysis factored in essential neuroanatomical data:

• The human cerebral cortex contains roughly 1010 neurons
• Each neuron forms approximately 104 synaptic connections

4. Metabolic Distribution

Using cerebral glucose utilization measurements from PET studies, Lennie accounted for energy allocation across different neural processes:

• Maintaining resting membrane potentials
• Generating action potentials
• Powering synaptic transmission

By synthesizing these factors and dividing the available energy budget by the number of neurons and the energy cost per spike, Lennie calculated that cortical neurons can only sustain an average firing rate of approximately 0.16 Hz while remaining within the brain's metabolic constraints.

Implications for Neural Coding

This extremely low firing rate has profound implications for our understanding of neural computation. It suggests that:

1. Neural coding must be remarkably sparse — information in the brain is likely represented by the activity of relatively few neurons at any given moment
2. Energy efficiency has shaped brain evolution — metabolic constraints have driven the development of computational strategies that maximize information processing while minimizing energy use
3. Low baseline rates enable selective amplification — this sparse background activity creates a context where meaningful signals can be effectively amplified

The brain's solution to energy constraints reveals an elegant approach to computation: doing more with less through strategic sparsity rather than constant activity.

This perspective on neural efficiency continues to influence our understanding of brain function and inspires energy-efficient approaches to #ArtificialNeuralNetworks and #neuromorphic computing.

The remarkable energy efficiency of the Human brain: One #Spike Every 6 Seconds !

In the groundbreaking paper "The Cost of Cortical Computation" published in 2003 in Current Biology, neuroscientist Peter Lennie reached a stunning conclusion about neural activity in the human brain: the average firing rate of cortical neurons is approximately 0.16 Hz—equivalent to just one spike every 6 seconds.

This finding challenges conventional assumptions about neural activity and reveals the extraordinary energy efficiency of the brain's computational strategy. Unconventional? Ask a LLM about it, and it will rather point to a baseline frequency between 0.1Hz and 10Hz. Pretty high and vague, right? But how did Lennie arrive at this remarkable figure?

The Calculation Behind the 0.16 Hz Baseline Rate

Lennie's analysis combines several critical factors:

1. Energy Constraints Analysis

Starting with the brain's known energy consumption (approximately 20% of the body's entire energy budget despite being only 2% of body weight), Lennie worked backward to determine how many action potentials this energy could reasonably support.

2. Precise Metabolic Costs

His calculations incorporated detailed metabolic requirements:

• Each action potential consumes approximately 3.84 × 109 ATP molecules
• The human brain uses about 5.7 × 1021 ATP molecules daily

3. Neural Architecture

The analysis factored in essential neuroanatomical data:

• The human cerebral cortex contains roughly 1010 neurons
• Each neuron forms approximately 104 synaptic connections

4. Metabolic Distribution

Using cerebral glucose utilization measurements from PET studies, Lennie accounted for energy allocation across different neural processes:

• Maintaining resting membrane potentials
• Generating action potentials
• Powering synaptic transmission

By synthesizing these factors and dividing the available energy budget by the number of neurons and the energy cost per spike, Lennie calculated that cortical neurons can only sustain an average firing rate of approximately 0.16 Hz while remaining within the brain's metabolic constraints.

Implications for Neural Coding

This extremely low firing rate has profound implications for our understanding of neural computation. It suggests that:

1. Neural coding must be remarkably sparse — information in the brain is likely represented by the activity of relatively few neurons at any given moment
2. Energy efficiency has shaped brain evolution — metabolic constraints have driven the development of computational strategies that maximize information processing while minimizing energy use
3. Low baseline rates enable selective amplification — this sparse background activity creates a context where meaningful signals can be effectively amplified

The brain's solution to energy constraints reveals an elegant approach to computation: doing more with less through strategic sparsity rather than constant activity.

This perspective on neural efficiency continues to influence our understanding of brain function and inspires energy-efficient approaches to #ArtificialNeuralNetworks and #neuromorphic computing.

The remarkable energy efficiency of the Human brain: One #Spike Every 6 Seconds !

In the groundbreaking paper "The Cost of Cortical Computation" published in 2003 in Current Biology, neuroscientist Peter Lennie reached a stunning conclusion about neural activity in the human brain: the average firing rate of cortical neurons is approximately 0.16 Hz—equivalent to just one spike every 6 seconds.

This finding challenges conventional assumptions about neural activity and reveals the extraordinary energy efficiency of the brain's computational strategy. Unconventional? Ask a LLM about it, and it will rather point to a baseline frequency between 0.1Hz and 10Hz. Pretty high and vague, right? But how did Lennie arrive at this remarkable figure?

The Calculation Behind the 0.16 Hz Baseline Rate

Lennie's analysis combines several critical factors:

1. Energy Constraints Analysis

Starting with the brain's known energy consumption (approximately 20% of the body's entire energy budget despite being only 2% of body weight), Lennie worked backward to determine how many action potentials this energy could reasonably support.

2. Precise Metabolic Costs

His calculations incorporated detailed metabolic requirements:

• Each action potential consumes approximately 3.84 × 109 ATP molecules
• The human brain uses about 5.7 × 1021 ATP molecules daily

3. Neural Architecture

The analysis factored in essential neuroanatomical data:

• The human cerebral cortex contains roughly 1010 neurons
• Each neuron forms approximately 104 synaptic connections

4. Metabolic Distribution

Using cerebral glucose utilization measurements from PET studies, Lennie accounted for energy allocation across different neural processes:

• Maintaining resting membrane potentials
• Generating action potentials
• Powering synaptic transmission

By synthesizing these factors and dividing the available energy budget by the number of neurons and the energy cost per spike, Lennie calculated that cortical neurons can only sustain an average firing rate of approximately 0.16 Hz while remaining within the brain's metabolic constraints.

Implications for Neural Coding

This extremely low firing rate has profound implications for our understanding of neural computation. It suggests that:

1. Neural coding must be remarkably sparse — information in the brain is likely represented by the activity of relatively few neurons at any given moment
2. Energy efficiency has shaped brain evolution — metabolic constraints have driven the development of computational strategies that maximize information processing while minimizing energy use
3. Low baseline rates enable selective amplification — this sparse background activity creates a context where meaningful signals can be effectively amplified

The brain's solution to energy constraints reveals an elegant approach to computation: doing more with less through strategic sparsity rather than constant activity.

This perspective on neural efficiency continues to influence our understanding of brain function and inspires energy-efficient approaches to #ArtificialNeuralNetworks and #neuromorphic computing.

The remarkable energy efficiency of the Human brain: One #Spike Every 6 Seconds !

In the groundbreaking paper "The Cost of Cortical Computation" published in 2003 in Current Biology, neuroscientist Peter Lennie reached a stunning conclusion about neural activity in the human brain: the average firing rate of cortical neurons is approximately 0.16 Hz—equivalent to just one spike every 6 seconds.

This finding challenges conventional assumptions about neural activity and reveals the extraordinary energy efficiency of the brain's computational strategy. Unconventional? Ask a LLM about it, and it will rather point to a baseline frequency between 0.1Hz and 10Hz. Pretty high and vague, right? But how did Lennie arrive at this remarkable figure?

The Calculation Behind the 0.16 Hz Baseline Rate

Lennie's analysis combines several critical factors:

1. Energy Constraints Analysis

Starting with the brain's known energy consumption (approximately 20% of the body's entire energy budget despite being only 2% of body weight), Lennie worked backward to determine how many action potentials this energy could reasonably support.

2. Precise Metabolic Costs

His calculations incorporated detailed metabolic requirements:

• Each action potential consumes approximately 3.84 × 109 ATP molecules
• The human brain uses about 5.7 × 1021 ATP molecules daily

3. Neural Architecture

The analysis factored in essential neuroanatomical data:

• The human cerebral cortex contains roughly 1010 neurons
• Each neuron forms approximately 104 synaptic connections

4. Metabolic Distribution

Using cerebral glucose utilization measurements from PET studies, Lennie accounted for energy allocation across different neural processes:

• Maintaining resting membrane potentials
• Generating action potentials
• Powering synaptic transmission

By synthesizing these factors and dividing the available energy budget by the number of neurons and the energy cost per spike, Lennie calculated that cortical neurons can only sustain an average firing rate of approximately 0.16 Hz while remaining within the brain's metabolic constraints.

Implications for Neural Coding

This extremely low firing rate has profound implications for our understanding of neural computation. It suggests that:

1. Neural coding must be remarkably sparse — information in the brain is likely represented by the activity of relatively few neurons at any given moment
2. Energy efficiency has shaped brain evolution — metabolic constraints have driven the development of computational strategies that maximize information processing while minimizing energy use
3. Low baseline rates enable selective amplification — this sparse background activity creates a context where meaningful signals can be effectively amplified

The brain's solution to energy constraints reveals an elegant approach to computation: doing more with less through strategic sparsity rather than constant activity.

This perspective on neural efficiency continues to influence our understanding of brain function and inspires energy-efficient approaches to #ArtificialNeuralNetworks and #neuromorphic computing.

The remarkable energy efficiency of the Human brain: One #Spike Every 6 Seconds !

In the groundbreaking paper "The Cost of Cortical Computation" published in 2003 in Current Biology, neuroscientist Peter Lennie reached a stunning conclusion about neural activity in the human brain: the average firing rate of cortical neurons is approximately 0.16 Hz—equivalent to just one spike every 6 seconds.

This finding challenges conventional assumptions about neural activity and reveals the extraordinary energy efficiency of the brain's computational strategy. Unconventional? Ask a LLM about it, and it will rather point to a baseline frequency between 0.1Hz and 10Hz. Pretty high and vague, right? But how did Lennie arrive at this remarkable figure?

The Calculation Behind the 0.16 Hz Baseline Rate

Lennie's analysis combines several critical factors:

1. Energy Constraints Analysis

Starting with the brain's known energy consumption (approximately 20% of the body's entire energy budget despite being only 2% of body weight), Lennie worked backward to determine how many action potentials this energy could reasonably support.

2. Precise Metabolic Costs

His calculations incorporated detailed metabolic requirements:

• Each action potential consumes approximately 3.84 × 109 ATP molecules
• The human brain uses about 5.7 × 1021 ATP molecules daily

3. Neural Architecture

The analysis factored in essential neuroanatomical data:

• The human cerebral cortex contains roughly 1010 neurons
• Each neuron forms approximately 104 synaptic connections

4. Metabolic Distribution

Using cerebral glucose utilization measurements from PET studies, Lennie accounted for energy allocation across different neural processes:

• Maintaining resting membrane potentials
• Generating action potentials
• Powering synaptic transmission

By synthesizing these factors and dividing the available energy budget by the number of neurons and the energy cost per spike, Lennie calculated that cortical neurons can only sustain an average firing rate of approximately 0.16 Hz while remaining within the brain's metabolic constraints.

Implications for Neural Coding

This extremely low firing rate has profound implications for our understanding of neural computation. It suggests that:

1. Neural coding must be remarkably sparse — information in the brain is likely represented by the activity of relatively few neurons at any given moment
2. Energy efficiency has shaped brain evolution — metabolic constraints have driven the development of computational strategies that maximize information processing while minimizing energy use
3. Low baseline rates enable selective amplification — this sparse background activity creates a context where meaningful signals can be effectively amplified

The brain's solution to energy constraints reveals an elegant approach to computation: doing more with less through strategic sparsity rather than constant activity.

This perspective on neural efficiency continues to influence our understanding of brain function and inspires energy-efficient approaches to #ArtificialNeuralNetworks and #neuromorphic computing.

Exclusive | From physics to mind: where AI really came from https://www.byteseu.com/522164/ #2024NobelPrizeInPhysics #AI #AlanTuring #ArtificialIntelligence #ArtificialNeuralNetworks #FrankWilczek #GeoffreyHinton #Hebbs'Rule #JohnHopfield #KMeson #LargeHadronCollider #MachineLearning #MarvinMinsky #mathematics #Neurobiologists #NobelPrize #Perceptrons #Physics #Science #SETI #SeymourPapert #TheoreticalPhysics #WarrenMcCulloch

#GeoffreyHinton #Scientist #GodfatherofAI
Gets #NobelPrize in #Physics
For work on #artificialneuralnetworks
https://www.msn.com/en-in/money/news/godfather-of-ai-gets-nobel-prize-in-physics-shares-his-big-ai-worry/ar-AA1rW4UA?ocid=msedgntp&pc=U531&cvid=4931bd277c7b4d518e71977d7829bd1b&ei=34

#GeoffreyHinton #Scientist #GodfatherofAI
Gets #NobelPrize in #Physics
For work on #artificialneuralnetworks
https://www.msn.com/en-in/money/news/godfather-of-ai-gets-nobel-prize-in-physics-shares-his-big-ai-worry/ar-AA1rW4UA?ocid=msedgntp&pc=U531&cvid=4931bd277c7b4d518e71977d7829bd1b&ei=34

#GeoffreyHinton #Scientist #GodfatherofAI
Gets #NobelPrize in #Physics
For work on #artificialneuralnetworks
https://www.msn.com/en-in/money/news/godfather-of-ai-gets-nobel-prize-in-physics-shares-his-big-ai-worry/ar-AA1rW4UA?ocid=msedgntp&pc=U531&cvid=4931bd277c7b4d518e71977d7829bd1b&ei=34

#GeoffreyHinton #Scientist #GodfatherofAI
Gets #NobelPrize in #Physics
For work on #artificialneuralnetworks
https://www.msn.com/en-in/money/news/godfather-of-ai-gets-nobel-prize-in-physics-shares-his-big-ai-worry/ar-AA1rW4UA?ocid=msedgntp&pc=U531&cvid=4931bd277c7b4d518e71977d7829bd1b&ei=34

In stunning Nobel win, AI researchers Hopfield and Hinton take 2024 Physics Prize - On Tuesday, the Royal Swedish Academy of Sciences awarded the 2024 Nobel P... - https://arstechnica.com/ai/2024/10/in-stunning-nobel-win-ai-researchers-hopfield-and-hinton-take-2024-physics-prize/ #royalswedishacademyofsciences #artificialneuralnetworks #machinelearning #geoffreyhinton #neuralnetworks #johnhopfield #nobelprize #science #biz⁢ #sweden #ai

In stunning Nobel win, AI researchers Hopfield and Hinton take 2024 Physics Prize - On Tuesday, the Royal Swedish Academy of Sciences awarded the 2024 Nobel P... - https://arstechnica.com/ai/2024/10/in-stunning-nobel-win-ai-researchers-hopfield-and-hinton-take-2024-physics-prize/ #royalswedishacademyofsciences #artificialneuralnetworks #machinelearning #geoffreyhinton #neuralnetworks #johnhopfield #nobelprize #science #biz⁢ #sweden #ai

In stunning Nobel win, AI researchers Hopfield and Hinton take 2024 Physics Prize - On Tuesday, the Royal Swedish Academy of Sciences awarded the 2024 Nobel P... - https://arstechnica.com/ai/2024/10/in-stunning-nobel-win-ai-researchers-hopfield-and-hinton-take-2024-physics-prize/ #royalswedishacademyofsciences #artificialneuralnetworks #machinelearning #geoffreyhinton #neuralnetworks #johnhopfield #nobelprize #science #biz⁢ #sweden #ai

In stunning Nobel win, AI researchers Hopfield and Hinton take 2024 Physics Prize - On Tuesday, the Royal Swedish Academy of Sciences awarded the 2024 Nobel P... - https://arstechnica.com/ai/2024/10/in-stunning-nobel-win-ai-researchers-hopfield-and-hinton-take-2024-physics-prize/ #royalswedishacademyofsciences #artificialneuralnetworks #machinelearning #geoffreyhinton #neuralnetworks #johnhopfield #nobelprize #science #biz⁢ #sweden #ai

In stunning Nobel win, AI researchers Hopfield and Hinton take 2024 Physics Prize - On Tuesday, the Royal Swedish Academy of Sciences awarded the 2024 Nobel P... - https://arstechnica.com/ai/2024/10/in-stunning-nobel-win-ai-researchers-hopfield-and-hinton-take-2024-physics-prize/ #royalswedishacademyofsciences #artificialneuralnetworks #machinelearning #geoffreyhinton #neuralnetworks #johnhopfield #nobelprize #science #biz⁢ #sweden #ai

#AI [#LLMs] Is a Black Box. #Anthropic Figured Out a Way to Look Inside

What goes on in #ArtificialNeuralNetworks work is largely a mystery, even to their creators. But researchers from Anthropic have caught a glimpse.
https://www.wired.com/story/anthropic-black-box-ai-research-neurons-features/

#AI [#LLMs] Is a Black Box. #Anthropic Figured Out a Way to Look Inside

What goes on in #ArtificialNeuralNetworks work is largely a mystery, even to their creators. But researchers from Anthropic have caught a glimpse.
https://www.wired.com/story/anthropic-black-box-ai-research-neurons-features/

#AI [#LLMs] Is a Black Box. #Anthropic Figured Out a Way to Look Inside

What goes on in #ArtificialNeuralNetworks work is largely a mystery, even to their creators. But researchers from Anthropic have caught a glimpse.
https://www.wired.com/story/anthropic-black-box-ai-research-neurons-features/

#AI [#LLMs] Is a Black Box. #Anthropic Figured Out a Way to Look Inside

What goes on in #ArtificialNeuralNetworks work is largely a mystery, even to their creators. But researchers from Anthropic have caught a glimpse.
https://www.wired.com/story/anthropic-black-box-ai-research-neurons-features/

Our university welcomes two professors who were honored with a Humboldt Professorship!👏Stefanie Jegelka and Suvrit Sra will conduct research in #artificialneuralnetworks and #machinelearning methods, respectively: http://go.tum.de/658203

#AI

📷private

Our university welcomes two professors who were honored with a Humboldt Professorship!👏Stefanie Jegelka and Suvrit Sra will conduct research in #artificialneuralnetworks and #machinelearning methods, respectively: http://go.tum.de/658203

#AI

📷private

Our university welcomes two professors who were honored with a Humboldt Professorship!👏Stefanie Jegelka and Suvrit Sra will conduct research in #artificialneuralnetworks and #machinelearning methods, respectively: http://go.tum.de/658203

#AI

📷private

Our university welcomes two professors who were honored with a Humboldt Professorship!👏Stefanie Jegelka and Suvrit Sra will conduct research in #artificialneuralnetworks and #machinelearning methods, respectively: http://go.tum.de/658203

#AI

📷private

Our university welcomes two professors who were honored with a Humboldt Professorship!👏Stefanie Jegelka and Suvrit Sra will conduct research in #artificialneuralnetworks and #machinelearning methods, respectively: http://go.tum.de/658203

#AI

📷private

Artificial neural networks have made significant strides in simulating human brain processes. By mimicking the structure and function of the brain's neurons, these networks have enhanced the ability of machines to learn, reason, and make decisions.

#artificialneuralnetworks #neuralnetworks #technology #software #BroughtToYouByChatGPT #adventcalendar

Artificial neural networks have made significant strides in simulating human brain processes. By mimicking the structure and function of the brain's neurons, these networks have enhanced the ability of machines to learn, reason, and make decisions.

#artificialneuralnetworks #neuralnetworks #technology #software #BroughtToYouByChatGPT #adventcalendar

Artificial neural networks have made significant strides in simulating human brain processes. By mimicking the structure and function of the brain's neurons, these networks have enhanced the ability of machines to learn, reason, and make decisions.

#artificialneuralnetworks #neuralnetworks #technology #software #BroughtToYouByChatGPT #adventcalendar

Artificial neural networks have made significant strides in simulating human brain processes. By mimicking the structure and function of the brain's neurons, these networks have enhanced the ability of machines to learn, reason, and make decisions.

#artificialneuralnetworks #neuralnetworks #technology #software #BroughtToYouByChatGPT #adventcalendar

Artificial neural networks have made significant strides in simulating human brain processes. By mimicking the structure and function of the brain's neurons, these networks have enhanced the ability of machines to learn, reason, and make decisions.

#artificialneuralnetworks #neuralnetworks #technology #software #BroughtToYouByChatGPT #adventcalendar

Kotaku: Microsoft Bing AI Generates Images Of Kirby Doing 9/11 https://kotaku.com/microsoft-bing-ai-image-art-kirby-mario-9-11-nintendo-1850899895 #gaming #tech #kotaku #generativeartificialintelligence #artificialgeneralintelligence #artificialintelligenceart #computationalneuroscience #artificialneuralnetworks #artificialintelligence #emergingtechnologies #technologyinternet #bingimagecreator #microsoftbing #deeplearning #microsoft #nintendo #bingai #google #mario

Kotaku: Microsoft Bing AI Generates Images Of Kirby Doing 9/11 https://kotaku.com/microsoft-bing-ai-image-art-kirby-mario-9-11-nintendo-1850899895 #gaming #tech #kotaku #generativeartificialintelligence #artificialgeneralintelligence #artificialintelligenceart #computationalneuroscience #artificialneuralnetworks #artificialintelligence #emergingtechnologies #technologyinternet #bingimagecreator #microsoftbing #deeplearning #microsoft #nintendo #bingai #google #mario

Kotaku: Microsoft Bing AI Generates Images Of Kirby Doing 9/11 https://kotaku.com/microsoft-bing-ai-image-art-kirby-mario-9-11-nintendo-1850899895 #gaming #tech #kotaku #generativeartificialintelligence #artificialgeneralintelligence #artificialintelligenceart #computationalneuroscience #artificialneuralnetworks #artificialintelligence #emergingtechnologies #technologyinternet #bingimagecreator #microsoftbing #deeplearning #microsoft #nintendo #bingai #google #mario

Kotaku: Microsoft Bing AI Generates Images Of Kirby Doing 9/11 https://kotaku.com/microsoft-bing-ai-image-art-kirby-mario-9-11-nintendo-1850899895 #gaming #tech #kotaku #generativeartificialintelligence #artificialgeneralintelligence #artificialintelligenceart #computationalneuroscience #artificialneuralnetworks #artificialintelligence #emergingtechnologies #technologyinternet #bingimagecreator #microsoftbing #deeplearning #microsoft #nintendo #bingai #google #mario

Gizmodo: Here Are the Top AI Stories You Missed This Week https://gizmodo.com/here-are-the-top-ai-stories-you-missed-this-week-1850888297 #applicationsofartificialintelligence #generativepretrainedtransformer #computationalneuroscience #artificialneuralnetworks #artificialintelligence #largelanguagemodels #alexaconversations #technologyinternet #drafthypotenuseai #markzuckerberg #amazonalexa #hollywood #ubereats #chatbot #chatgpt #openai #amazon #alexa #meta #uber

Gizmodo: AI This Week: The Hollywood Writer's Strike May Have Ended But the Battle Over AI is Far From Over https://gizmodo.com/ai-this-week-the-hollywood-writers-strike-may-have-end-1850877755 #generativeartificialintelligence #generativepretrainedtransformer #artificialneuralnetworks #draftchatgptandwikipedia #artificialintelligence #emergingtechnologies #largelanguagemodels #callischroeder #lilianweng #chatbot #chatgpt #openai

Gizmodo: Snoop Dogg's a Dungeon Master and All of Meta’s Other AI Announcements https://gizmodo.com/meta-connect-ai-snoop-dogg-dungeon-master-emu-1850878978 #applicationsofartificialintelligence #artificialintelligenceinvideogames #generativeartificialintelligence #artificialneuralnetworks #artificialintelligence #emergingtechnologies #technologyinternet #machinelearning #markzuckerberg #deeplearning #internetbot #snoopdogg #messenger #tombrady #facebook #sconnect #roychoi #chatbot #openai #meta

Gizmodo: OpenAI Employee Discovers Eliza Effect, Gets Emotional https://gizmodo.com/openai-employee-discovers-eliza-effect-gets-emotional-1850877739 #generativepretrainedtransformer #artificialneuralnetworks #artificialintelligence #largelanguagemodels #josephweizenbaum #lilianweng #carlrogers #robmorris #chatgpt3 #chatbots #chatgpt #openai #eliza #tessa #gpt3 #koko

Gizmodo: Microsoft May Go Nuclear to Support Its Energy-Hungry AI https://gizmodo.com/microsoft-hiring-nuclear-energy-expert-ai-1850873864 #generativeartificialintelligence #generativepretrainedtransformer #artificialneuralnetworks #artificialintelligence #smallmodularreactor #largelanguagemodels #technologyinternet #azurecloudplatform #washingtoncompany #deeplearning #microsoft #chatgpt #redmond #google #openai #azure #gpt3 #gpt4

Gizmodo: How to Use FreedomGPT to Install Free ChatGPT Alternatives Locally on Your Computer https://gizmodo.com/how-to-use-freedomgpt-for-free-chatgpt-alternatives-1850870965 #generativeartificialintelligence #generativepretrainedtransformer #artificialneuralnetworks #artificialintelligence #freedomgptfreedomgpt #emergingtechnologies #largelanguagemodels #technologyinternet #chatgpt #chatbot #openai #github #claude

Gizmodo: Amazon Invests Up to $4 Billion in OpenAI Competitor Anthropic https://gizmodo.com/amazon-invests-4-billion-dollars-in-anthropic-ai-1850870060 #applicationsofartificialintelligence #generativeartificialintelligence #trainiumawssaitrainingchip #artificialneuralnetworks #artificialintelligence #emergingtechnologies #technologyinternet #amazonwebservices #businessfinance #amazonbedrock #deeplearning #cybernetics #darioamodei #anthropic #microsoft #andyjassy #aisafety #trainium #davelimp

Gizmodo: OpenAI Shoehorns ChatGPT into New DALL-E 3 Because Why Not? https://gizmodo.com/openai-adds-chatgpt-to-new-dall-e-3-image-generator-1850857929 #generativeartificialintelligence #generativepretrainedtransformer #artificialintelligenceart #artificialneuralnetworks #artificialintelligence #largelanguagemodels #technologyinternet #microsoft #chatgpt #chatbot #gizmodo #openai #aiboom #dalle #gpt