levels, reduction vs. emergence. any science - theories at each of several levels because most...
TRANSCRIPT
• Any science - theories at each of several levels because most complex systems observed - constructed in a hierarchy of levels
• Interaction between elements at any single level - described without specifying any but very general properties of the elements at the next level below, and without considering dynamics at the next level above.
• Ex - A car: Mechanisms, Sub-mechanisms, quantum level (Herbet Simon 2007)
Levels• Explanation of cognition in micro- (neural
states) and macro-levels (mental phenomena)
• 2 types of interactions:- Intralevel (horizontal) interactions -
components at same level - Interlevel (vertical) interactions - the
relationship between levels• “Levels”: ontological, organizational
(mechanisms or systems), epistemological or description or analysis.
David Marr - Three Levels of Description• Computational level- What information is computed and why- What the system is capable of doing- Deep Blue and Kasparov – equivalent
(Dawson 2007) • Representation and algorithm (software)- What program is used- What are the symbols, how are processed- Deep Blue and Kasparov - different• Hardware
Ex: Linguistic understanding
• Task: Identify syntax and meaning corresponding to speech sounds.
• Algorithm: What kind of computation and
mental representations?
• Implementation: Which part of the brain?
• Ontological levels - radical E
• Organizational levels – modest E
• Levels of analysis - specific value E
- Analytical levels partially depend upon
viewing nature as organized into parts andwholes. (McCauley 07)
3-levels versus more-levels• Understanding cognition = Understanding brain
at different levels = Levels of organization (Gordon Shepherd)
Levels of the brain:• Whole brain • Large systems and pathways in the brain (e.g.
sensory pathways)• Properties of specific centers + local circuitsproperties of neurons (single cell recording)structures within neurons (dendrites/axons)• Individual synapses + molecular properties of
membranes and ion channels (Dawson 2007)
• Using principles of organization and scale, Churchland and Sejnowski [1992]
7 sub-levels within neuroscience • molecules • synapses • neurons• networks • maps • sub-systems • central nervous system overall
• Ontologically different levels + radical E → Dualism – Rejected ↔ Anomalies
→
• Humans = Limited knowledge (McGinn 89)
• Distinction ontology-epistemology (levels of analysis and epistemic emergence - weak and strong)
• Organizational levels (related to layered view of nature)
Smolensky (88) - Connectionism• 3 levels of analysis: conceptual (symbolic,
subconceptual, neuronal)• Relationship between sub-symbolic-
symbolic levels (similar to relationship between micro-macro):
- Conceptual phenomena – necessary conceptual (symbolic) level = Macro-description of cognition
- Sub-conceptual phenomena – necessary sub-conceptual level = micro-description of
cognition
• “Mental Rs + Processes: Not supported by the same formal entities
• Nets - 2 levels:
1) Formal, algorithmic specification of processing mechanisms
2) Semantic interpretation
→ Must be done at 2 levels of description (Smolensky 1991)
• “1 nivel: Procesele mentale reprezentate de nivelul de descriere numerice a unit-lor, legat-lor, ecuatii de evolutie activarilor (NU interpretare semantica) (Smolensky)
• “2 nivel: Activitati la nivel larg permit interpretari, dar patt-le astfel fixate nu sunt descrieri clare a procesarii”
• “Metrica semantica a sistemului impune o similaritate pt. continut acolo unde exista o similaritate pt. vehicul (=patter-ri similare).” (Clark)
• Each hidden unit = Microfeature
• Microfeature=Unintelligible (‘interpretation’
depends upon its context = other microf-s simultaneously present)
• A collection of microfeatures (number of different hidden units) can represent a concept that could be represented by a symbol in a symbolic model
• Symbolic account of a network is only an approximate account (Dawson 2007)
The ‘‘Massive Redeployment Hypothesis’’
(Horst 2007, p. 164)
• Localism - meta-analysis of over one hundred fMRI studies by Michael Anderson (forthcoming)
• Neural correlates of performance of some cognitive task and identified, through subtraction analysis, areas of brain that were differentially active during task
• Authors: Regions identified were ‘‘memory regions,’’ ‘‘attention regions,’’ depending on nature of task studied
• Anderson:
(a) Most regions studied - utilized in multiple tasks, and indeed multiple types of tasks (e.g., attention and memory)
(b) Most tasks involved multiple Brodmann areas (see Table 8.2).
→ ‘‘Cognitive tasks’’ not stand in a one-to-one relation with Brodmann areas, but in a many-to-many relationship (Figure 8.3).
↔ A ‘‘massive redeployment’’ of preexisting brain areas to obtain new functionality.
• Anderson: Evolutionary history, brain areas redeployed (originally modular units)
• Evolution - a new strategy for acquiring new functionality = Redeploying existing functionality in ensembles of neural areas working together (Not genetical mutations)
2. Reduction (R) vs. Emergence (E)• The history of E[1] complicated, many
interpretations[2] • E = Vertical relationships low-level + high-level
properties• Reduction or emergence - what is reduced or
emerge to what: property or level
[1] Kim: “Since around 1990, the idea of emergence has been making a big comeback, from decades of general neglect and disdain - analytic philosophy.”
[2] Kim: “‘Emergence’ is very much a term of philosophical trade; it can pretty much mean whatever you want to mean…”
Reduction (R) (van Gulick 2001)
Ontological R (objects, properties, events ...)
• elimination
• identity
• composition
• supervenience
• realization
Epistemological R (concepts, theories, models, frameworks)
• Replacement
• Theoretical–Derivational (Logical Empiricist)
• A priori Conceptual Necessitation
• Expressive Equivalence
• Teleo–Pragmatic Equivalence
Ernest Nagel [‘61/’79] + Logical empiricists• Reduction of scientific theories + whole
sciences• Logical derivation + bridge principles• Constraint (1): “Derivability” of reducedtheory from reducing theory• Constraint (2): “Connectability”
• “Homogeneous” R of a theory and “heterogeneous” reduction of a theory or a science
Wisatt (76) • “Intralevel” relations = Relations over time
between successive theories in some science
vs. • “Inter level” relations = Cross-scientific relations
between theories that reign at the same time at different analytical levels in science
→ Methodological + ontological implications for theories and sciences contrast
vs.• Anti-R: Multiple realizability + irreducibility of
conscious experience (McCauley 07)
• van Gulick: E = “’Xs are more than just Ys’ and that ‘Xs are something over and above Ys.’”
• E features - beyond the features of parts from which they emerge = “metaphysical E” (it refers to the relation between real things) or “epistemic E” (cognitive explanatory relations about real world items). (van Gulick 2001)
• Ontological and epistemological E - people many times conflate them. (Silberstein and McGeever 1999; O’Connor and Wong 2002)
• Some properties = combinations of parts at same level.
• E properties - different from ∑ parts → Novelty• Crane: novel properties of object- “determinable
properties whose determinates are not had by all of the object’s parts.”
Ex: Surface colour and wetness
Or E properties of whole - supervenient properties of parts
• Such properties = “over and above" physical properties
2.1 Ontological emergence
(1) Specific value E = Whole and parts - features of same kind, different specific subtypes/values
(Ex: A bronze statue + molecular parts - common property = Mass)
(2) Modest kind E: “whole - features different in kind from those of its parts …” (Ex: Color, life)
(3) Radical kind E: “The whole - features
a) Different in kind from parts
b) Of kind whose nature and existence is not necessitated by the features of its pars + macro-laws influence micro-laws/entities (van Gulick)
• Ontological E - world = Layered view of nature
• Ontologically emergent properties are not determined or reducible to basic properties.(ex. QM)
• Ontological E - Controversial
• Strong E property = “High-level phenomenon arises from a low-level domain, but truths concerning that phenomenon - not deducible even in principle from truths in low-level domain.” (Chalmers 2006)
• If strong E phenomena - Not deducible from laws of physics ↔ New laws of nature for
consciousness
• Colorblind scientist + zombies
• Consciousness - supervenes on neural states
Epistemological E
• Epistemic E - Incapacity explain/predict property of whole system in terms of its parts
• Property of whole - determined by properties of parts
• Epistemic E - Weak and strong E (Gulick)
• Property = Epistemological E - determined to/deducible from intrinsic properties of fundamental entities that compose objects
• Difficult to explain/predict such a property in terms of its fundamental constituents
• “Epistemologically E properties - novel at level of description”
• High-level phenomenon = Weak E to low-level when that phenomenon is “unexpected” in accord with laws from low-level (Chalmers 2006)
• “Unexpected” - E properties - somehow deductible from low-level properties
• Ex: “Game of life”, connectionist networks, evolution (for intelligent creatures), high-level patterns CA.
• Weak E
(a) High-level properties of system are not of any of its parts
(b) Deductibility without reducibility
O’Connor and Wong (2002)• Predictive: E properties = Features of
complex systems - not predicted despite knowledge of features + laws of parts
• Irreducible-Pattern: E properties + laws = Features of complex systems governed by true, lawlike generalizations within a special science
• = Irreducible to fundamental physical theory for conceptual reasons
• Macroscopic patterns - Not captured in concepts + laws physics
Stephan (2002, 1998): Weak, synchronic, and diachronic E
• Weak E: Properties of system - E if they belong to the system as a whole, but not to the parts of that system ↔ Property reductionism
• System as a whole=∑parts + organization
Ex.: Connect. nets + self-organization and artificial intelligence
• Diachronic E: Novelty + unpredictability of the system that evolves
• Difference weak-diachronic E = Unpredictability of properties
• Difference diachronic-synchronic E = Irreducibility of properties
• If one property/entity not existed before and suddenly comes into existence ↔ Diachronically new
• Synchronic novelty is time independent• These 2 irreducibilities → Downward
causation or epiphenomenalism
Irreducible property (a) Does not follow from the behavior of the
system’s parts that has this property (b) Does not follow from the behavior of the
system’s parts in constellations simpler than the system
• Ex: Qualia = Synchronic E properties
• Nonreductive physicalism (Davidson, 1970; Putnam, 1972; 1978; Fodor, 1974; Boyd, 1980; Searle, 1992; Van Gulick, 1992) (in Gulick 2001)
vs. • Left (dualists - Chalmers, 1996; Hasker 1999)
and right (reductive physicalists such as Kim, 1989)
• Fodor – “autonomy of the special sciences” vs.• Old unity of science view (Oppenheim and
Putnam, 1958): all true theories must ultimately be translatable into language of physics - rejected
CURRENT OPTIONS ON MIND/BODY PB.• Reduction: 10 versions (see Figure 3)• Emergence: 10 versions (see Figure 7)• Other Options:• Mainstream: Nonreductive physicalism• More radical: Fundamental Dualism- Property- Substance• Pan (Proto-) Psychism• Dual Aspect Monism (Spinoza, Strawson)• Multi-Revolutions View (Penrose, McGinn)