Private
Public Access
0
0

conductor(generic_systems_fields): Phase 3 OCR - 33 frames OCR'd via winsdk in 1.9s

This commit is contained in:
2026-06-21 23:21:11 -04:00
parent 3c4dd5c20f
commit d1d98c85ce
3 changed files with 472 additions and 0 deletions
@@ -0,0 +1,468 @@
# OCR Results
## frame_00001.jpg
```
O
How diverse is intelligence?
What are the limits?
How do we find out?
M. Levin, 10.1002/aisy.202401034
```
## frame_00002.jpg
```
"Intelligence is a fixed goal with variable means of achieving it."
— William James
Does any goal count?
Are any means allowed?
Does anything fall outside this definition?
```
## frame_00003.jpg
```
Biological Theory
https•J/doiorg/10.1007/s13752Q505234
ORIGINAL ARTICLE
Mind Everywhere: A Framework for Conceptualizing Goal-
Directedness in Biology and Other Domains—Part One
Michael Levin12(tY • David B. Resnik
Received: 31 March 2025 / Accepted: 22 November 2025
@Konrad Lorenz Institute for Evolution and Cognition Research 2025
Abstract
What makes a system—evolved, engineered, or hybrid—describable by teleological and mentalistic terms such as intel-
ligent, goal-directed, cognitive, and intentional? In this two-part article, we review classical thought on teleology in the
life sciences and defend a new approach to goal-directedness that stems from an emerging field—diverse intelligence. This
field seeks to characterize what all active agents, regardless of their composition or provenance, have in common. Our
approach emphasizes: (l) empirical testability (not philosophical commitments to linguistic categories). (2) fecundity in
discovery of new capabilities (not just reductive mechanistic explanations of results after they are made, but worldviews
that facilitate and enable novel research), (3) operationalization of terminology by reference to conceptual and empiri-
cal toolkits shown to be effective for a given system (cognitive and teleological claims are really hypotheses of optimal
interaction protocols), and (4) continuity of human goal-directedness with our unicellular origins (which implies a need
for models of scaling of cognition). In Part One, we review historical and contemporary debates about teleology in biol-
```
## frame_00004.jpg
```
Physics of Life Reviews 47 (2023) 35-62
Contents lists available at ScienceDirect
Physics of Life Reviews
journal homepage: www.elsevier.com/locate/plrev
ELSEVIER
Review
Path integrals, particular kinds, and strange things
a.b.c.*
Dalton A.R. Sakthivadivel e
Karl Friston a-c, Lancelot Da Costa
Conor Heinsc f g, Grigorios A. Pavliotis b, Maxwell Ramstead LC, Thomas Parra
The free energy principle (FEP) describes a simple relationship between the dynamics of a
random dynamical system and a description of its behaviour as engaging in inference. The FEP
originated in neuroscience as an attempt to describe brain function and behaviour (Friston et al.
2006) and has since been extended to describe several kinds of things in the biological and
physical realms (Friston 2013; Friston et al. 2021) through a special kind of mechanics—a
Bayesian mechanics—that shares the same foundations with quantum, statistical, and classical
mechanics (Friston 2019; Friston et al. 2022). This paper is part of a series of technical papers
describing the FEP in progressively simpler and more qualified terms (Friston, 2013; Friston,
2019; Friston et al., 2022).
```
## frame_00005.jpg
```
Inert particles
with no acuve states
Active particles
with active states
Conservative particles
with classical dynanucs
Strange particles
with hidden active states
External states
s
Sensory states
s
a
q, (11)
Active states
Internal states
Ihi - v p
A causal sink .
violates Newton's
3rd Law
Fig. 2
```
## frame_00006.jpg
```
Physics first, intuitions later ...
How do we guarantee that this interaction
respects all physical symmetries?
We want the generic case, that describes
any interacting systems, regardless of
scale or structure.
```
## frame_00007.jpg
```
Answer: Start with the generic case
U, an isolated system
(no environment,
no interaction)
The internal dynamics "PU must respect Newton's 3rd Law everywhere.
No sources, no sinks — because U has no environment.
QU conserves momentum, energy, information (so is unitary).
```
## frame_00008.jpg
```
An isolated (no environment) U is a productive assumption.
This gives us conservation of information, so unitarity, so linearity.
Hence is a linear operator on the state space , a Hilbert space.
In background time t, we can write TU(t) = exp((-i/h)Hu(t)), where
h is an action and HU is a Hamiltonian (energy) linear operator on YLJ.
A finite value of h 4-+ no singularities.
This is quantum theory (QT). "Isolation is all you need."
```
## frame_00009.jpg
```
Because everything in sight is linear, we can do a linear decomposition.
Boundary "B
U =AÄ, HU = HA + HÄ + HAÄ. HAÄ is the interaction between A and Ä.
This is the generic, symmetry-preserving interaction we wanted.
```
## frame_00010.jpg
```
Without loss of generality,
HAÄ = kBTkEINßk(oz, zki)
where k = A or Ä, Oz is a
z-spin operator, and
System A
zki is a local z reference
frame.
This makes "B an N-qubit
holographic screen.
4
Prepare
Measure
(Oz, z 1)
Measure
Prepare
Prepare
Measure
Prepare
Measure
Measure
(Oz, z 2)
Prepare
Measure
(Oz, z N)
Prepare
System Ä
(Oz, Z
Boundary
```
## frame_00011.jpg
```
HAÄ tells of how A and Ä act on each other.
We want it to tell us how they influence each other.
These are the same if but only if A and Ä have conditionally-
independent states.
So we need to require that II...J> = IAÄ> = This is state
separability = absence of entanglement.
Separability requires weak (or sparse) coupling. Formally, the
dimension of HAÄ is small: dim(HAÄ) << dim(HA), dim(HÄ).
Intuitively, the evolutions of A and Ä are almost independent.
```
## frame_00012.jpg
```
Minimal physics .....................+ FEP
If A and Ä are separable, dim(HAÄ) << dim(HA), dim(HÄ):
• HAÄ fully describes information exchange between A and Ä;
• The boundary "B functions as a Markov Blanket;
• Variational free energy (VFE) measures interaction strength;
• Minimizing VFE is keeping HAÄ weak while allowing
thermodynamic exchange;
• Predictability = constrained interaction.
A and Ä maintain their identities as distinct systems only while
their boundary "B remains intact — no rips, no explosions!
```
## frame_00013.jpg
```
"Intelligence is a fixed goal with variable means of achieving it."
— William James
Does any goal count?
>> There's always at least one: continuing to exist as an entity.
Are any means allowed?
>> Whatever the internal dynamics HA and HÄ are capable of.
Does anything fall outside this definition?
>> No, it's completely generic.
```
## frame_00014.jpg
```
We have a generic, symmetry-preserving description.
It is consistent with and even explains the FEP.
But is the behavior that counts as "intelligent" interesting?
Are there limits on what kinds of systems can exhibit interesting
behavior?
How do we find out?
```
## frame_00015.jpg
```
What is interesting behavior?
• Surprising, unpredictable in practice
• Only approximately predictable (only predictable if coarse-grained)
• Unpredictable in principle
• Learns from experience
• Memory-dependent
• Context-dependent
• Distributions of outcome values violate Kolmogorov, outcome
probabilities undefined in principle
```
## frame_00016.jpg
```
Operationally,
State transition probabilities derived from finite observations
do not converge to predictive adequacy.
Induction from finite data doesn't work.
19th Century "mechanical" expectations are violated.
We know Life violates them. What else does?
```
## frame_00017.jpg
```
Hint: Moore's theorem (1956):
Finite input-output experiments cannot uniquely determine the
"machine table" (internal state-transition probabilities) of a generic
classical Black Box.
Example: Box with an internal clock, e.g. time bomb.
```
## frame_00018.jpg
```
Hint: Conway-Kochen "free will" theorem (2006, 2009):
Special relativity and quantum theory together rule out local
(past light cone) determinism.
"If experimenters make choices, electrons do too."
```
## frame_00019.jpg
```
Hint: QT from singularity removal (Tipler, 2014):
The simplest formal removal of singularities from classical
physics reproduces Bohm's "quantum potential."
(N. Gisin: Newton-Laplace physics wasn't singular because
it wasn't local. Einstein introduced strict locality.)
```
## frame_00020.jpg
```
These all suggest:
Generic systems (can) display interesting behavior.
How do we make this precise?
How do we understand it?
How can we use it to explain and/or predict?
```
## frame_00021.jpg
```
The setting:
A's measurements
and action choices
are computed by HA
(In FEP, A's GM).
Observations
Boundary 'B
and actions (1/0)
live here, on "B.
dim(HAÄ) << dim(HA), dim(HÄ)
Inputs and outputs are much less complex than the computations that
generate them.
```
## frame_00022.jpg
```
This dimensionality/complexity difference immediately tells us:
Recurrence of I'B> does not imply recurrence of IÄ>.
Behavior generically depends on "hidden" internal states, i.e. on
Ä's memory or internal context.
We can represent this formally as an internal "geometric" or Berry phase.
Chris Fuchs: all physical systems have "interiority."
```
## frame_00023.jpg
```
Geometric phase changes are
introduced by transports along
curves in state spaces. These
are "holonomy" operations.
E.g. Blattner (2026):
1
7
2
3
6
5
4
Hidden regenerative state in planarians:
A geometric model of bioelectric memory using Tangential
Action Spaces
Nlarcel Blattner
```
## frame_00024.jpg
```
Why is this important?
Non-trivial geometric phase dependence — non-trivial holonomy — is
a sufficient resource for universal quantum computation.
It constructs a map: I'B> 1B IÄ' > for arbitrary I'B'>.
• Zanardi, P. and Rasetti, M. Holonomic quantum computation. Phys. Lett.
A 264 (1999), 94-99.
• Pachos, J. and Zanardi, P. Quantum holonomies for quantum computing.
Int. J. Mod. Phys. B 15 (2001), 1257-1286.
```
## frame_00025.jpg
```
What is physically implemented computation?
Input ....................+ Output
Input -.......................+ Output
"P(t) implements fon Input if and only if these diagrams commute.
The "interpretation" is a projection/inverse embedding: = 8-1.
```
## frame_00026.jpg
```
Why is this important?
Embeddings are injective: one to many.
Polycomputation is generic.
Indeed, managing thermodynamic flow requires that "informative"
sector projections are proper samples of We never look at
everything the computer is doing.
```
## frame_00027.jpg
```
We can think of computation as "scattering" in data-structure space
(CF et al., 2509.19772).
An ideal classical computer implementing an algorithm for f looks like:
"Ready" state
Input
Output
"Ready" state
This is a useful coarse-graining, but the observed "Ready state"
(a proper projection of does not pick out a unique machine state
IÄ>. This is Moore's Theorem from 1956, updated.
```
## frame_00028.jpg
```
Even a classical OS
accumulates internal
state changes as it
executes.
"Side projects" are
inevitable in generic
systems.
' 'Ready"
State
Input 1
Input 2
Input N
Output 1
Output 2
Output N
Final
state
```
## frame_00029.jpg
```
We can also represent geometric phase changes as reference frame
changes:
(Oz, )
Holonomy
If Ä exhibits non-trivial holonomy, how it acts on 'B will change.
```
## frame_00030.jpg
```
Why is this important?
(Oz, ) and (Oz, don't commute! (Oz, )( Oz,
Non-commuting QRFs generate non-causal context dependence.
In this case, joint probability distributions on observational
outcomes are undefined (violate Kolmogorov).
Fields and Glazebrook, Int. J. Theor. Phys. 62 (2023) 159
```
## frame_00031.jpg
```
All of these "interesting" kinds of behavior are generic!
• Surprising, unpredictable in practice
• Only approximately predictable (only predictable if coarse-grained)
• Unpredictable in principle
• Learns from experience
• Memory-dependent
• Context-dependent
• Distributions of outcome values violate Kolmogorov, outcome
probabilities undefined in principle
They all result from separability: big systems with small boundaries.
```
## frame_00032.jpg
```
O
QT provides a precise,
general, strongly empirically
validated foundation for
Diverse Intelligence.
It tells us that intelligence
and persistent observability
go hand in hand.
M. Levin, 10.1002/aisy.202401034
```
## frame_00033.jpg
```
Thank you
Questions?
Thanks to Jim Glazebrook.
```
@@ -0,0 +1,2 @@
Phase 2 Keyframes for C:\projects\manual_slop\conductor\tracks\video_analysis_generic_systems_fields_20260621\artifacts\video.mp4
OK: kept 33 frames
@@ -0,0 +1,2 @@
Phase 3 OCR for C:\projects\manual_slop\conductor\tracks\video_analysis_generic_systems_fields_20260621\artifacts\frames (winsdk)
OK: OCR'd 33 frames in 1.9s