LANGUAGE, COGNITION, AND MUSICAL
EMOTIONS
University of Birmingham
School of Computer Science
23 Nov 2010
Leonid Perlovsky
Harvard University and the AF Research Lab
http://leonid-perlovsky.com/
OUTLINE
• Mathematical difficulties
– Computational Complexity
• Dynamic Logic (DL)
– Examples: recognition, situations
– The Knowledge Instinct (KI)
– Model of PSS
• Future research
–
–
–
–
The mind hierarchy, emotion of the beautiful
Language and Cognition
Musical emotions: cognitive function, origin
Experimental tests: psychological, computer science
MATHEMATICAL DIFFICULTIES:
COMPUTATIONAL COMPLEXITY
• Perception, cognition, language involve evaluating
large numbers of combinations
– Pixels -> objects -> scenes -> abstract concepts
– Sounds -> words -> phrases -> …
• Combinatorial Complexity (CC)
– A general problem (since the 1950s)
• Detection, recognition, tracking… language…
• Pattern recognition, neural networks, rule systems…
• Combinations of 100 elements are 100100
– This number > the size of the Universe
• > all the events in the Universe during its entire life
• Related to Gödelian limitations of logic
DYNAMIC LOGIC (DL)
• DL unifies formal and fuzzy logic
– A process-logic “from fuzzy to crisp”
• Maximizes similarity between models and
signals
• Overcomes CC -> fast algorithms
• Proven in neuroimaging experiments (Bar, 2006)
– Initial representations-memories are vague-fuzzy
– “close-eyes” experiment
EXAMPLE: RECOGNITION
DL “from vague to crisp”
a
b
c
d
e
f
g
h
Signal / Clutter ratio ~ 100 times improvement
ARISTOTLE VS. GÖDEL
• Aristotle
– Logic: a supreme way of argument
– Forms: representations in the mind
 Form-as-potentiality evolves into form-as-actuality
 Potentialities are not logical -> logical actualities, (Dynamic Logic)
– Language and thinking are closely linked
• From Boole to Russell: formalization of logic
– Logicians eliminated from logic uncertainty of language
– Hilbert: formalize rules of mathematical proofs forever
• Gödel (the 1930s)
– Logic is not consistent
 Any statement can be proved true and false
• Aristotle and Alexander the Great
objects
EXAMPLE: Learning Situations
DATA (SORTED)
Situations (sorted)
objects
SITUATIONS: DATA (RANDOM)
Situations (random)
SITUATION LEARNING: ERRORS
THE MIND, KNOWLEDGE INSTINCT,
AND DL
• Mechanisms of the mind:
– Instincts are like sensors measuring vital signals
– Emotions are neural signals connecting instinctual and decisionmaking mechanisms
– Concepts are mental representations, models
– Behavior
– Hierarchy
• The knowledge instinct (KI)
–
–
–
–
–
Concept-models always have to be adapted
KI increases similarity between models and the world
DL is mathematically equivalent to KI
KI-emotions: (dis)harmony between concepts and the world
KI-emotions are aesthetic emotions
 In every act of perception and cognition
DL and Perceptual Symbol System
(PSS)
• PSS
– Mental representations are distributed bits and pieces of
experiences
– Simulations are the reenactment of perceptual experiences
– Simulations reassemble perceptual experiences
– Simulations are the mechanisms of perception and cognition,
symbol-processes
• DL
– DL process “from vague-to-crisp” models PSS simulations
– DL is a general mechanism of interacting bottom-up and topdown signals
– Applicable to all levels of the cognitive processing hierarchy
– How amodal signals emerge in PSS (final crisp state)
OUTLINE
• Mathematical difficulties
– Computational Complexity
• Dynamic Logic (DL)
– Examples: recognition, situations
– The Knowledge Instinct (KI)
– Model of PSS
• Future research
–
–
–
–
The mind hierarchy, emotion of the beautiful
Language and Cognition
Musical emotions: cognitive function, origin
Experimental tests: psychological, computer science
HIERARCHY OF COGNITION
COGNITION
…
abstract ideas
Top representations
SURROUNDING
LANGUAGE
Purpose:
unify entire life experience
LANGUAGE
aesthetic emotions are felt as
beautiful
Conceptual
abstractand emotional contents
are not differentiated
words/phrases
…
…
language
descriptions
Vague and unconscious
of abstract
representations
thoughts
situations
Purpose: unify lower level
phrases
representations
aesthetic emotions are conscious
phrases for situations
objects
Vague and less conscious
representations
words
aesthetic emotions could be
conscious
words for objects
Sensory-motor
signals
Concrete conscious perceptions
language
Sensory-motor
aesthetic emotions are below
sounds
language consciousness
models
INTEGRATED
LANGUAGE AND COGNITION
• How language and cognition interact
– Words and objects: zillions of combinations, how do we learn correct
ones?
– Each concept (mental representation) has linguistic and cognitive dual
model
Mm = { Mmcognitive, Mmlanguage };
– Language and cognition are fused at vague pre-conceptual level
• before words and concepts are learned
• Arbib: “language prewired brain”
• Language and cognition mechanisms
–
–
–
–
In a newborn mind, concept-models are dual vague blobs
Language is learned “ready-made” from surrounding language
Language models guide development of cognitive model (situations)
Cognitive concepts are learned to match language models
INTEGRATED HIERARCHIES
• High level cognition is only possible due to language
COGNITION
…
abstract ideas
situations
LANGUAGE
…
abstract
words/phrases
SURROUNDING
LANGUAGE
…
language
descriptions
of abstract
thoughts
phrases
phrases for situations
objects
words
words for objects
Sensory-motor
signals
Sensory-motor
language models
language
sounds
RECOGNITION
• 2007 Gabor Award
- The top engineering award from International Neural
Network Society (INNS)
• 2007 John L. McLucas Award
- The top scientific award from the US Air Force
• 2000 Best Paper Award, Zvezda
• Elected to the Board of Governors of INNS
• On Editorial Boards of 6 journals
BIBLE, NOBEL PRIZE, LANGUAGE
• Why Adam was expelled from paradise ?
– Did not want to think (cognitive models are vague)
– Choose ready-made language rules (Maimonides, 13th c.)
• Nobel Prize 2002, Kahneman (and Tversky)
– Decisions are basically irrational
– Not KI, but rules-heuristics (crisp language models)
– KI-cortex (OFC) vs. rules-amygdala (DL & LP)
• Language vs. irrational-rules
– Language contains wealth of cultural knowledge, rules
– Remember: opened eyes hide vague mental images
– Similarly language hide vague abstract concepts
MUSIC IS A MYSTERY
• Aristotle, Kant, Darwin
- “must be ranked amongst the most mysterious (abilities) with which (man)
is endowed”
• Contemporary psychologists
– Pinker, Justus, Hustler, McDermott, Houser, Trainor, Fitch, Huron, Cross,
Mithen, Juslin, a series of reviews in Nature…
– Dissanayake: “aberrance of modernity…”
– Masataka: “Music is a human cultural universal that serves no obvious adaptive purpose,
making its evolution a puzzle for evolutionary biologists”
16-Sep-05
18
THEORETICAL HYPOTHESIS
• Language
– split concepts and emotions
– accelerated differentiation of knowledge
• Diverse knowledge induces contradictions in psyche
- Cognitive dissonances
- Need to connect language to cognition
• Music restores unity of psyche
- Along with diversity of knowledge
• A number of contradictions is very large
- Two pieces of knowledge contain a contradiction, cognitive dissonance
- Combinations of knowledge pieces are very large, this is the reason for “infinity” of
musical emotions
• Music creates the diversity of emotions necessary to continue culture
16-Sep-05
19
ORIGIN
• Animals
- Unity of psyche
• Human voice separated into language & music
- Language reduced emotionality and increased semantics
- Violated unity of psyche
- Diverse knowledge brings advantage if combined with unity of
psyche
- To restore unity, part of voice became more emotional, less
semantic
• Music
- Evolved together with language to keep synthesis of differentiated
psyche
- “It is difficult to keep the scissor blades together”
16-Sep-05
20
EVOLUTION
• Undifferentiated proto language-music
• Motherese
• Earliest musical instruments 9,000 years old (may be older)
• Every major cultural-spiritual advantage required new music
– Jaynes J., The origin of consciousness, 1976
– Weiss P, Taruskin R. Music in the western world, 1984
– Cultural process is the combination of oppositions, differentiation and synthesis,
music is needed to keep psyche together
• Few examples
– Contemporary consciousness (4-5 BCE) Thales, Zechariah, Confucius –
antiphony
– Human emotions - Renaissance (13-16) – tonality
– The highest contradiction - Reformation (16) – Baroque, fugue, Bach
– Rationality – Enlightenment (17-18) – Classicism (Bach sons, Mozart)
– Self – (19) – Romanticism (Beethoven, Chopin, Tchaikovsky)
– Unconscious+objective – (20) all mixed-up, Modern, Postmodern
– Pop songs – restore unity of everyday consciousness
– Rap – a contemporary dithyramb, connects conscious and unconscious
16-Sep-05
21
EXPERIMENTAL TESTS
• There is no words for musical emotions, so how can we
measure them? What to measure?
• Psychologists
– (1)Measure subjective emotional similarities (or distances) among
musical phrases
– Mathematically evaluate dimensionality of the emotional space
– (2)Measure emotional similarities among musical phrases and
cognitive dissonances
– (3)Measure fMRI or EEG signatures of musical phrases
• Computer scientists
– (1)Collect context (50 words) around emotional words on Google
– Measure similarities (word overlap) between emotional words
– Mathematically evaluate dimensionality of the emotional space
– (2)Repeat for cognitive dissonances (look for word “choice”)
16-Sep-05
22
PUBLICATIONS
380 publications
3 books
OXFORD UNIVERSITY PRESS
(2001; 3rd printing)
Neurodynamics of High Cognitive
Functions
with Prof. Kozma, Springer, 2007
Sapient Systems
with Prof. Mayorga, Springer, 2007
2011:
Dynamic Logic, Springer
FUTURE DIRECTIONS
research, predictions and testing
• Mathematical development
KI in the hierarchy, combine with language and emotions
Multi-agent simulations of cultural evolution
• Psycholinguistic experiments
Measure emotionality of various languages in labs
• Music: theoretical and experimental
Direct effect on emotions
Concurrent evolution of music, consciousness, and cultures
Multi-agent simulations of cultural evolution
• Brain imaging
Brain regions used by languages, music, in different cultures
Neural mechanisms connecting language and cognition
• Semantic Web and Cyberspace
Adaptive ontologies
Learn from human users, acquire cultural knowledge
Enable culturally-sensitive communication
Help us understand each other and ourselves
Multi-agent simulations of cultural evolution
• Improve human condition and understanding around the globe
Develop predictive cultural models, integrate spiritual and material causes
Identify language and music effects that can advance consciousness and reduce tensions
24
BACK-UP
•
•
•
•
•
•
Structure of the mind
Neural Modeling Fields
Dynamic logic
Neuro-imaging experimental confirmation
Beautiful and sublime
Cultural models
16-Sep-05
25
STRUCTURE OF THE MIND
• Concepts
– Models of objects, their relations, and situations
– Evolved to satisfy instincts
• Instincts
– Internal sensors (e.g. sugar level in blood)
• Emotions
– Neural signals connecting instincts and concepts
• e.g. a hungry person sees food all around
• Behavior
– Models of goals (desires) and muscle-movement…
• Hierarchy
– Concept-models and behavior-models are organized in a “loose”
hierarchy
NEURAL MODELING FIELDS
from signals to concepts
• Bottom-up signals
– Pixels or samples (from sensor or retina)
x(n), n = 1,…,N
• Top-down concept-models
Mm(Sm,n), parameters Sm, m = 1, …;
– Models predict expected signals from objects
• The knowledge instinct = maximize similarity
between signals and models
L = l ({x}) =   l (x(n) | Mm)
n
m
– MN items: all associations of pixels and models (=>CC)
– New mathematical technique, DL, overcame this difficulty
DYNAMIC LOGIC (DL)
non-combinatorial max of knowledge
• Start with a set of signals and unknown
object-models
– any parameter values Sm
– associate models with signals (vague)
– (1) f(m|n) = r(m) l (n|m) /  r(m') l (n|m')
m'
• Improve parameter estimation
– (2) Sm = Sm + a  f(m|n) [ln l (n|m)/Mm]*[Mm/Sm]
n
• Continue iterations (1)-(2). Theorem: MF is a
converging system (from vague to crisp)
- similarity increases on each iteration
- aesthetic emotion is positive during learning
DL AND BRAIN IMAGING
• Neuro-imaging experiments proved that the brain works as
predicted by dynamic logic
• Bar et al (2006), Harvard University proved
- Bottom-up signals (from eye retina) interact with top-down signals (from
memory-models)
- Initial top-down signals are vague
- These interactions are unconscious
• Barsalou et al (2006), Emory University proved
- Distributed vague representations in the mind
16-Sep-05
29
BEAUTIFUL AND SUBLIME
• The highest aesthetic emotion, beautiful
– improvement of the highest models (at the top of the hierarchy)
– feel emotion of beautiful
• Beautiful “reminds” us of our purposiveness
– the “top” model unifies all our knowledge
– vague
– we perceive it as our purpose (“aimless purposiveness”)
• Beauty is separate from sex
– sex uses all our abilities, including beauty
• Religiously sublime is related to behavior
MODELS OF
CULTURAL EVOLUTION
• Differentiation, D, synthesis, S,
hierarchy, H
dD/dt = a D G(S);
dS/dt = -bD + dH
H
= H0 + e*t
G(S) = (S - S0) exp(-(S-S0) / S1)
INTERACTING CULTURES
• Two cultures
– dynamic and traditional
– slow exchange by D and S
dDk/dt = ak Dk G(Sk) + xkDk
dSk/dt = -bkDk + dkHk + ykSk
Hk
= H0k + ek*t
CULTURE AND LANGUAGE
• Culture is transmitted through language
• Examine mechanisms of
– Language and cognition
– Language and emotion
• The following are hypotheses in need of
verification
EVOLUTION OF CULTURES
• The knowledge instinct
- Two mechanisms: differentiation and synthesis
• Differentiation
- At every level of the hierarchy: more detailed concepts
- Separates concepts from emotions
• Synthesis
- Connects concepts and emotions (knowledge and life)
 Connects language and cognition
 Created in the hierarchy: concepts acquire meaning at the next level
• Evolutionary dynamics
- Complex interaction of opposing mechanisms
16-Sep-05
34
EMOTIONAL
SAPIR-WHORF HYPOTHESIS
• Language affects thinking and behavior
– Bhartrihari 5th CE (India), Humboldt 1836, Nietzsche 1876
– Benjamin Whorf and Edward Sapir in the 1930s
 E.g., people better perceive colors, which have words in their language
• Recent history
– “Out of favor.” Chomsky separated language and cognition
– Recent resurgence of interest
• We have to understand cultural differences
– “European” thinking is not the only way
• Emotional differences are no less
important than semantical
EMOTIONS IN
INTEGRATED HIERARCHIES
• Look top-down: differentiation, more detailed concept-models, less emotions
• Look bottom-up: synthesis, unifying general models, more emotions
• At the top: meaning and purpose, emotions of the beautiful and sublime
Meanings
vague
emotional
COGNITION
…
SURROUNDING
LANGUAGE
LANGUAGE
…
abstract ideas
situations
…
abstract
words/phrases
language
descriptions
of abstract
thoughts
Objects
crisp
conceptual
hierarchy:
conceptual
phrases
phrases for situations
objects
throughout
Synthesis
emotions words
crisp
low
words for objects
Sensory-motor
signals
Sensory-motor
language models
language
sounds
emotions
EMOTIONS IN LANGUAGES
• Animals
- Undifferentiated concepts-emotions-behaviors-vocalization
- Vocal tract is controlled from ancient emotional limbic system
• Human language evolution
-
Language evolved toward semantics and less emotions
Still emotions are needed, otherwise, no synthesis, no meaning
Two emotional centers: limbic (involuntary) and cortex (conscious)
Emotionality: in voice sound (melody of speech)
• Emotional differences among languages
-
All languages evolved toward less emotionality
More semantic flexibility, but potential to lose meanings
“Too fast” evolution => lose meaning
“Too slow” evolution => culture stagnates
Speed is determined by grammar, by inflections (Humboldt’s
firmness)
16-Sep-05
37
LANGUAGE
EMOTIONS AND CULTURES
• Conceptual content of culture: words, phrases
Easily borrowed among cultures
• Emotional content of culture
In voice sound (melody of speech)
Determined by grammar
Cannot be borrowed among cultures
• English language (Diff. > Synthesis)
Weak connection between conceptual and emotional (since 15 c)
Pragmatic, high culture, but may lead to crises (lost meaning)
• Arabic language (Synthesis > Diff.)
Strong connection between conceptual and emotional
Cultural immobility, but strong feel of identity and purpose
16-Sep-05
38
DYNAMIC CULTURE
Average synthesis, high differentiation; oscillating solution
Knowledge accumulates; no stability
TRADITIONAL CULTURE
High synthesis, low differentiation; stable solution
Stagnation, stability increases
INTERACTING CULTURES
1) Early: Dynamic culture affects traditional culture, no reciprocity
2) Later: 2 dynamic cultures stabilize each other
Knowledge accumulation + stability
Descargar

Document