Presentation is loading. Please wait.

Presentation is loading. Please wait.

Design principles for adaptive self-organizing systems

Similar presentations

Presentation on theme: "Design principles for adaptive self-organizing systems"— Presentation transcript:

1 Design principles for adaptive self-organizing systems
Finding Fluid Form Symposium University of Brighton December 9-10, 2005 Design principles for adaptive self-organizing systems Peter Cariani Department of Physiology Tufts Medical School Boston

2 My trajectory Organismic biology MIT mid 1970s) Biological cybernetics & epistemology (1980s) Biological alternatives to symbolic AI Howard Pattee, Systems Science, SUNY-Binghamton Temporal coding of pitch & timbre (1990s) Auditory neurophysiology, neurocomputation How is information represented in brains? Commonalities of coding across modality & phyla Neural timing nets for temporal processing Auditory scene analysis Possibilities inherent in time codes Temporal alternatives to connectionism signal multiplexing; adaptive signal creation broadcast

3 Evolution of ideas Elaboration of structures & functions over time in biological, social, and technological realms, What makes new functions possible (functional emergence)? Can we put these principles to work for us? Is structural complexification by itself sufficient? (No) Notions of function & functional emergence are needed. What kinds of functions? Sensing, effecting, coordinating Is pure computation on symbols sufficient? (No) How are brains/minds capable of open-ended creativity? Neural codes, temporal codes, timing nets Neural coding of pitch in the auditory system Rethinking the architecture of the brain: Temporal alternatives to connectionism Adaptive signal creation & multiplexing, Broadcast coordinative strategies

4 Combinatoric vs. creative emergence

5 An example Exhaustive description Limited description
All permutations of single digits consisting of 6 tokens All permutations of 6 arbitrarily defined objects One well-defined set having 610 permutations BOUNDED Ill-defined number of sets, each w. 610 permutations UNBOUNDED

6 Describing the world: Two perspectives
Omniscent “God’s eye view” Postulational, ontological analytical mode Perspective of the limited observer epistemological empirical mode Appearance of new structures over time Violations of expectations “Surprise”

7 Well-defined vs. ill-defined realms
Exhaustive description God’s eye view Limited description Limited observer System-environment as well-defined realm Environment as ill-defined realm Description is dependent on set of observables (environment has as many properties as one can measure) Description of all-possible organism-environment relations CLOSED WORLD ASSUMPTION OPEN WORLD ASSUMPTION No fundamental novelty is possible All novelty is combinatoric Combinatoric and Creative emergence


9 Philosophy Ontology Aristotelian hylomorphism Material substrate that exists independently of us, yet whose form is largely ill-defined, incompletely known Organization is embedded in material system (e.g. mind is the organization of the nervous system) Conscious awareness requires a particular kind of regenerative informational organization embedded in a material system (cybernetic functionalism) Aristotle's Causes: Multiple complementary modes of explanation that answer different kinds of questions


11 Philosophy Epistemology Pragmatism (truth of a model related to its purpose) Perspective of the limited observer Relativism: different observational frames & purposes Analytical, empirical and pragmatic truths Analytic: truths of convention (non-material truths, finist mathematics) Empirical: truths of measurement, observation (science) Pragmatic: truths of efficacy & aesthetics (engineering, art) Constructivism & epistemic autonomy: by semi-freely choosing our own observables & concepts, we construct ourselves (for better or worse)

12 adaptive, self-organizing systems
Design principles for adaptive, self-organizing systems We are interested in designing & fabricating systems that autonomously organize themselves to elaborate structures & improve functions in response to challenges of their environments in ways that are meaningful and useful to us and/or them

13 Design principles for adaptive, self-organizing systems
Richness of material possibility (e.g. polymeric combinatorics) + Ability to steer & stabilize structure (feedback to structure: sensors, coordination mechanisms, effectors) + Means to interact w. material world (sensing, action = "situatedness", semantics) + Means to evaluate actions re: purposes (goal-laden representations, "intentionality") => Material system capable of adaptive, elaboration & improvement of informational functions

14 Design principles for adaptive, self-organizing systems
Richness of material possibility (need polymers, replicated aperiodic structure, Schrodinger's aperiodic crystal, analog dynamics, ill-defined interactions) Ability to steer & stabilize structure (need controls on self-production of internal structure, enzymes) Means to interact w. material world (Need sensors, effectors, neural nets) Means to evaluate actions re: purposes (Need natural selection or internal goal states, limbic system)

15 Vibratory dynamics of matter
Cymatics: Bringing Matter to Life with Sound Hans Jenny Richness of material possibility Complexity is easy Steerable complexity is hard

16 Two phases in creative learning processes
Design principles for adaptive, self-organizing systems VARIATION + SELECTION + INHERITANCE => ADAPTATION Material possibility+ Steer, stabilize, specify, inherit + Sensorimotor interaction + Evaluation => ASOS Two phases in creative learning processes Expansive phase: generation of possibility Realm of free & open creation e.g. scientific imagination and hypothesis creation Contractive phase: selection of best possibilities Realm of clarity & rigorous evaluation e.g. hypothesis testing (clarity, removal of ambiguity)

17 Analog dynamics and discrete symbols
We will also argue that one almost inevitably needs mixed analog-digital systems for complex systems: i.e. systems w. analog dynamics constrained by digital states ("symbols") for reliable replication of function for inheritability of adaptive improvements Analog and digital are complementary modes of description analog descriptions - continuous differential equations digital descriptions - discrete states & ST rules/probabilities Digital states or discrete symbols are ultrastable basins of attraction

18 Different theoretical approaches to understanding brains and their functions
Dynamical systems approaches Neural information processing Symbol- processing differential growth homeostasis analog representations processing states & switches branching discrete

19 Requisite: sensorimotor loops Inner and outer loops
metabolism: self-production steering: percept-action coordinations action perception interaction w. environment

20 Von Uexküll’s umwelts

21 McCulloch’s internal and external loops


23 Self-conscious description of the modeling process: Hertzian modeling relation: measurement & computation

24 The choice of observables
Finding the variables The would-be model maker is now in the extremely common situation of facing some incompletely defined "system," that he proposes to study through a study of "its variables." Then comes the problem: of the infinity of variables available in this universe, which subset shall he take? What methods can he use for selecting them? W. Ross Ashby, "Analysis of the system to be modeled" in: The Process of Model-Building in the Behavioral Sciences, Ohio State Press, pp ; reprinted in Conant, ed. Mechanisms of Intelligence

25 Choice of primitive features for classifiers
The choice of observables - analogous problems Choice of primitive features for classifiers Evolution of sensory organs in organisms Choice of sensors for robots

26 Semiotics of adaptive devices
Feedback to state Feedback to structure alters functionalities

27 Semiotic relations (Charles Morris)

28 Evaluate re: goals Frontal & limbic systems Internally generated pattern sequences sensory systems motor systems

29 Adaptivity in percept-action loops (Cariani)

30 Pure computation (state-determined system, no independent informational transactions w. environment)

31 Fixed robotic device Fixed sensors, coordinators, and effectors; Purely reactive and driven by its inputs; Incapable of learning

32 Computationally adaptive device
Trainable machines Neural networks Adaptive classifiers Genetic algorithms Robots w. adaptive programs Capable of learning new percept-action mappings (classifications)

33 Some observations about adaptability
Whatever functionalities are fixed, the designer must specify works for well-defined problems & solutions advantage: predictable, reliable behavior drawback: problems of specification Whatever is made adaptive must undergo a learning phase needed for ill-defined problems & solutions some unpredictability of solutions found creative behavior! the more autonomy, the more potentially creative Consequently, there are tradeoffs between adaptability & efficiency autonomy/creativity & control/predictability

34 Evolution/adaptive construction of new sensors
sensory evolution immune systems perceptual learning capable of learning new perceptual categories new feature primitives (new observables)

35 Epistemic autonomy • When a system can choose its own categories – through which it perceives and acts on the world – that system achieves some limited degree of epistemic autonomy. • A rudimentary electrochemical device was built by cyberneticist Gordon Pask in 1958 that grew its own sensors to create its own “relevance criteria.”

36 "With this ability to make or select proper filters on its inputs, such a device explains the central problem of epistemology. The riddles of stimulus equivalence or of local circuit action in the brain remain only as parochial problems." Warren McCulloch, preface ,Gordon Pask (1961) An Approach to Cybernetics.

37 Principles of action/use
1. Front-ends for trainable classifiers Useful in ill-defined situations where one does not a priori know what features are adequate to effect a classification 2. Adaptive, self-organizing sensors Grow structures over analog-VLSI electrode arrays in order to sense new aspects of the world. Use biochemical and/or biological systems coupled to an electrode array 3. Materially-based generator of new behaviors (adaptive pattern-generators) Similar steerable, ill-defined systems could be used to generate new patterns (sound, images) in an open-ended way that is not at all obvious to the observer/controller 4. Epistemic autonomy Device chooses how it will be connected to the outside world; what aspects of the material world (categories) are relevant to it. (Symbol grounding, frame problem)

38 Feedback to state vs. feedback to structure
A thermostat is limited in the information that it can gain from its environment by the fixed nature of its sensors. It has feedback to state, but not feedback to structure. The amount of information that such a system can extract from its environment is finite at any time, and bounded by its fixed structure. A system capable of sensory evolution or perceptual learning has the ability to change its relation to its environs. Such a system has an open-ended set of observational primitives. It has both feedback to state and feedback to structure. The amount of information that such a system can extract from its environment is finite at any time, but unbounded. Such a system is open-ended.

39 Analog dynamics without inheritable constraint (Hans Jenny)

40 von Neumann's kinematic (robotic) self-reproducing automaton (1948)

41 Inheritable construction analog dynamics constrained & selected by discrete symbols
Purely analog adaptive system must be trained each generation Genetic algorithm + Pattern grammar for guiding construction constrained search Symbolically-encoded memory permits results of an optimization process to be passed to subsequent generations

42 The homeostat


44 Relation to Ashby's homeostat
Analog sensor/controller Uniselector 25x25x25x25 = 390k construction possibilities -> variety of the control system, unconstrained search Evaluation of ability to control inputs

45 Relation to Ashby's homeostat
Analog sensor/controller Uniselector 25x25x25x25 = 390k construction possibilities -> variety of the control system, unconstrained search Evaluation of ability to control inputs

46 (ill-defined structure)
Ashby's homeostat Analog controller (ill-defined structure) Adaptive analog controller Structure of particular controllers is unknown to designer Requisite variety for control is the number of alternative controllers available 25x25x25 = 390,625 Uniselector evaluate (in bounds?) Environment

47 The homeostat & the brain
A few cybernetics-inspired accounts of brain function Sommerhoff (1974) Logic of the Living Brain Klopf, The Selfish Neuron Arbib, The Metaphorical Brain Most successful neuroscientific application of cybernetics: W.Reichardt's analysis of fly optomotor loop The homeostat never caught on as a brain metaphor Some possible reasons: Homeostats never were cast in terms of neural nets No obvious digital uniselector function in the brain Predominance of problems of pattern recognition and formulation of coherent action over simple problems of internal regulation

48 The brain as an adaptive self-organizing system
Ideas that flow from cybernetics and theoretical biology: Brains as signal self-production systems related to reverberant loops (a la Lorente, Lashley, Hebb, McCulloch, Pitts & many others) 2) Brains as pattern-resonance systems related to Lashley, Hebb, many others 3) Brains as multiplexed signaling and storage systems holographic paradigms, Longuet-Higgins, Pribram,John 4) Brains as mass-dynamics, broadcast systems 5) Brains as communications nets that create new signals 6) Brains as temporally-coded pulse pattern systems I believe all this is possible using temporal pattern codes.

49 Regeneration of parts

50 Von Neumann’s kinematic self-reproducing automaton

51 Autopoiesis and autocatalysis

52 Symbolically-guided self-production

53 Autopoiesis and autocatalysis Life is built upon cycles of self-production

54 Brain function may be based on self-productions of spike patterns Hebbian reverberant eigenstates and regenerative temporal patterns McCulloch & Pitts (1943) Nets with circles render activity independent of time and semi-autonomous re: the environment von Foerster (1948) brain eigenstates as a form of ST memory

55 Why the mind is in the head Warren McCulloch
L.A. Jeffress, ed. Cerebral Mechanisms of Behavior (The Hixon Symposium, Wiley, 1951, reprinted in Embodiments of Mind, MIT, 1965, concluding lines) This brings us back to what I believe is the answer to the question: Why is the mind in the head? Because there, and only there, are hosts of possible connections to be formed as time and circumstance demand. Each new connection serves to set the stage for others yet to come and better fitted to adapt us to the world, for through the cortex pass the greatest inverse feedbacks whose function is the purposive life of the human intellect. The joy of creating ideals, new and eternal, in and of a world, old and temporal, robots have it not. For this my Mother bore me.

56 The brain as a self-regenerating pattern-resonance system

57 Tuning in nervous systems Minds as pattern-resonances
The same [resonance] is true of all bodies which can yield notes. Tumblers resound when a piano is played, on the striking of certain notes, and so do window panes. Nor is the phenomenon without analogy in different provinces. Take a dog that answers to the name "Nero." He lies under your table. You speak of Domitian, Vespasian, and Marcus Aurelius Antonius, you call upon all the Roman Emperors that occur to you, but the dog does not stir, although a slight tremor of his ear tells you of a faint response of his consciousness. But the moment you call "Nero" he jumps joyfully towards you. The tuning fork is like your dog. It answers to the name A. Ernst Mach, Popular Lectures, “The fibers of Corti” c. 1865

58 Pattern resonances: neural assemblies emitting annotative tag signals that elaborate a regenerating signal pattern

59 Phase-locking in visual neurons
Temporal pattern codes Phase-locking in visual neurons (Horseshoe crab ommatidium, 5-15 Hz flashes) Phase-locking in auditoryl neurons Cat auditory nerve fibers, 250 Hz tone Javel,

60 Phase-locking in auditory nerve fibers
250 Hz tone

61 Frequency and time in the auditory nerve

62 Phase-locking of discharges in the auditory nerve
Cat, 60 dB SPL

63 Temporal coding in the auditory nerve
Work with Bertrand Delgutte Cariani & Delgutte (1996) Dial-anesthetized cats. 100 presentations/fiber 60 dB SPL Population-interval distributions are compiled by summing together intervals from all auditory nerve fibers. The most common intervals present in the auditory nerve are invariably related to the pitches heard at the fundamentals of harmonic complexes.

64 Phase-locking in visual thalamus (LGN) Stimuli: Drifting sinusoidal gratings

65 Color vision

66 Temporal coding of taste



• Temporal sieves • Extract (embedded) similarities • Multiply autocorrelations RECURRENT TIMING NETS • Build up pattern invariances • Detect periodic patterns • Separate auditory objects

70 Potential advantages of temporal pattern pulse codes & timing nets
Multiplexed signal transmission Orthogonality of patterns; less interference Flexible multimodal integration Encoding of signal identity in itself (logical type) Liberate signals from wires Broadcast of signals + selective reception Nonlocal computational operations Mass action (statistical representations) Open-ended creation of new signal primitives

71 Temporal expectancies in perception
Music, brain, and time In the image of the digital computer, we conceptualize brains as distributed logic machines. However, temporal correlation machines may prove to be a better metaphor. Temporal expectancies in perception Temporal patterning of body processes Temporal structure of movement Temporal expectations and reward structure (dopamine system, conditioning) Temporal memory traces Music may have the profound effects that it does because 1) it directly impresses its temporal structure on the activity of many neuronal populations, and 2) the neural codes & computations underlying experience are inherently temporal. Andy Partridge, xtc

72 Conclusions Design principles for self-organizing systems Structural complexity alone is not sufficient Pure computation alone is not sufficient Requisites Sensors & effectors Mixed digital-analog design Feedback to structure, self-production Inheritable, replicable (digital) plans Combinatorics of digital strings Rich analog, ill-defined dynamics Goal states and steering/selection mechanisms Possibility of brain as temporally-coded self- organizing system

73 Temporal coding of sensory information

74 From cochlea to cortex Primary auditory cortex (Auditory forebrain)
Auditory thalamus Inferior colliculus (Auditory midbrain) Lateral lemniscus Auditory brainstem Auditory nerve (VIII) Cochlea


76 Phase-locking to a 300 Hz pure tone
Period histogram (1100 Hz) First-order interval histogram (1500 Hz) # spikes Evans, 1982

77 Auditory nerve

78 Vowel Formant Regions


80 Codes are defined in terms of their functional roles
What spike train messages have the same meanings? (functional equivalence classes) What constitutes a difference that makes a difference? Temporal codes are neural codes in which timings of spikes relative to each other are essential to their interpretation.

81 Neural resonances


83 Phase-locking of an LGN unit to a drifting sinusoidal grating
Temporal modulation frequency PST Histograms Interval Histograms 4 Hz 8 Hz 16 Hz 32 Hz 64 Hz

84 Adaptive systems Adaptation ~ adjustment Sensing ~ measurement
• Depending upon the self-modification process, adaptive systems change in different ways. • They become tuned to their environments, on the percept on the action side internally: anticipating events, forecasting effects • New sensors create new linkages with the external world new perceptual primitives new observables new modes of adjustment • New effectors create new modes of action

85 Switching between reverberant states


87 Frequency ranges of (tonal) musical instruments
10k 8 6 5 4 3 2 1 0.5 0.25 Frequency ranges of (tonal) musical instruments 27 Hz 4 kHz 262 Hz 440 880 110

88 Measurement and tuning
mediates interactions with external world permitting behavior contingent upon perception Adaptive systems that create their own measurements are possible (we may be such systems) Tuning involves adjustment of internal relations to external relations, i.e. adaptive resonance It is possible to envision brains and minds as resonant systems that operate on patterns rather than coupled via energetic relations

89 Areas of self-modifying media
Self-modifying computers Coevolution between humans and computers Emergent human-machine couplings Pask’s Conversation theory Computers need means of independently accessing the world and creating their own concepts (epistemic autonomy) Self-organizing materials Electrochemical Ferromagnetic Biological-silicon interfaces Intelligent materials Mixed digital-analog feedback systems

90 Phase-locking of an LGN unit to a drifting sinusoidal grating
Temporal modulation frequency PST Histograms Interval Histograms 4 Hz 8 Hz 16 Hz 32 Hz 64 Hz

91 Phase-locking in visual thalamus (LGN) Stimulus: Drifting sinusoidal gratings


• Temporal sieves • Extract (embedded) similarities • Multiply autocorrelations RECURRENT TIMING NETS • Build up pattern invariances • Detect periodic patterns • Separate auditory objects

94 Build-up and separation of two auditory objects
Two vowels with different fundamental frequencies (F0s) are added together and passed through the simple recurrent timing net. The two patterns build up In the delay loops that have recurrence times that correspond to their periods. Period = 10 ms Vowel [ae] F0 = 100 Hz Vowel [er] F0 = 125 Hz Period = 8 ms

95 Contingent vs. logically-necessary “truths”
Sensing vs computing Contingent vs. logically-necessary “truths” Methodological issues: What distinguishes sensing from other kinds of informational operations? A sensing process must be contingent, it must have two or more possible outcomes to reduce uncertainty, whereas A computation (formal operation) must be logically-determined, it must always produce the same outcome given the same initial state

96 Computers and brains •Digital computers presently are capable of recombination-based creativity, but do not presently create new primitives for themselves. • Brains, on the other hand, are self-modifying systems with rich analog dynamics that can serve as substrates for formation of new informational primitives. • Contemplation of self-modifying systems is essential if we are to construct artificial systems that can create meaning for themselves. • We need such systems when problems are ill-defined, or when we desire open-ended creative possibilities.

97 Neural resonances

98 Overview I: Measurement in adaptive systems
• We discuss the semiotics and functional organization of different adaptive systems. • Adaptive systems reorganize their internal structure in order to improve their performance. • We consider how systems with sensors, effectors, and coordinative faculties can adaptively modify their internal structures and functions. • We consider how this adaptivity leads to emergent functions and behaviors.

99 Overview IV: Creativity, autonomy, and specification
• Creativity has two levels: Recombination of existing primitives De novo creation of new kinds of primitives • Inherent tradeoffs: Specifiability vs. autonomy Predictability/reliability vs. creativity

100 Homeostat

101 Grey Walter's device

102 Conceptions of “emergence”
• Appearance of new structures, functions, behaviors • Novelty that was not predictable from what came before Varieties • Structural emergence (appearance of new structures, org. levels) • Computational emergence (unexpected results) • Thermodynamic emergence (dissipative systems) • Functional emergence (flight, color vision) • Emergence-relative-to-a-model (perspectivist, operationalist)

103 Methodological issues
How can we identify the existence of information processing operations in artificial and natural systems? How can we distinguish measurement, computation, and effector operations from each other in an unknown material system? How can we detect changes in these functionalities, such that we know that our devices or organisms have modified them adaptively? We need operational distinctions. We need to be able to parse a state-transition graph.

104 Recognizing determinate & contingent events

105 State-transitions and observer-operations
How do we distinguish measurements and computations (such that we can also detect changes in system behavior)?


107 Emergence relative to an observer: What does the observer have to do to his/her own model to continue successfully predicting the material system’s behavior?

108 Evolution of observer's model

109 Opening up the sensory interface: Break-out strategies for creating new observables
1) construction of new sensors 2) modification of existing sensors 3) interposition of sensory prostheses 4) active measurements 5) creation of new internal sensors

110 Prosthesis: augmentation of functionalities
All technology is prosthesis.

111 Operational states and procedures in a scientific model
Explicate realm of symbols (well-defined) Implicate realm of material process (ill-defined)

112 Active measurement

113 Neural assemblies as internal sensors

Download ppt "Design principles for adaptive self-organizing systems"

Similar presentations

Ads by Google