From UBC Wiki
Jump to: navigation, search

Key Concepts and Vocabulary


Abjads (consonant alphabets): Symbols represent consonants only, or consonants+vowel. Examples include: Hebrew, Arabic, Syriac.

Abugidas (syllabic alphabets): Symbols represent consonants and vowels (syllables). Consonants have an inherent vowel, which can be changed or muted by use of diacritics and other modifications.

Alphabets: Sets of letters, usually in a set order, which represent one or more phonemes.

Syllabaries: Phonetic writing system that consists of symbols that represent syllables. Differs from abugidas in that symbols cannot be split into consonants and vowels.

Semanto-phonetic writing systems: System with symbols that represent both sound and meaning.

Digraph: two characters that represent a single sound such as [ea] in "pear" or [ai] in "pair"

Graphemes: Contrastive symbol. Two symbols are contrastive if the replacement of one by the other causes a different pronunciation or word.

Allographs: non-constrastive symbol such as <a> <A> <a>

Polyphone: single grapheme that represents two sounds. Example: <x> in . "xylophone" where it's a /ks/ sound.

Mora: Measure of syllabic weight. It can be vowel + onset, vowel, coda. A syllable can have more than one mora.


Cognitive Psychology:

Cognitive Neuropsychology:

Cognitive Neuroscience:



Computer Science

Modules and Class Notes

Module One: Intro to COGS

7 September: Welcome

The assessment for the course will consist in:

Participation: 8% After each class (starting next week) online reflection on reading and lecture Individual Assignments: 22% Due at the ends of modules one and three Final Exam: 25% 18 December, 7-10pm. Midterm Group Exercise: 10% In class, 12 October; Open books; No tech. Final Project Draft Proposal: 10% End of class on 2 November Final Written Project and Presentation: 25% On day of poster session, 28 November, report compiled in UBC Wiki. The people teaching the course are:

Dr. Ryan Taylor (Office Hours: Monday, 9:30-10:30, Stores Road Annex, ISRL)

Dr. Rebecca Reh (Office Hours: Monday 3:00-4:30, Kenny, 2408)

Dr. Christopher Mole (Office Hours: Wednesday, 9:00-10:00, or by appointment, Buchanan E wing, 369)

Andy Moldwin (Office Hours: Friday 11:00-12:00, SCRF 1022)

Daniel Fredericks (Office Hours: Tuesday 11:00-12:00, SCRF 1023)

The TAs can be contacted with administrative queries at Substantive questions are best asked in person at an office hour.

Five Core Courses Cogs 200 Cogs 300 Cogs 303 Cogs 401 Cogs 402

Two Great Mysteries

  • How is it possible for a material creature to display the properties that are essential to life. (self replicate)
  • How is it possible for a material creature to display the properties that are essential to the mind.

12 September: Linguistics

Different types of writing systems

  • Abjads: Represent consonants only, or consonants + some vowels. Mostly written right to left. Full vowel sounds are usually not there, but can be included through diacritics.
  • Abugidas: Symbols for consonants and vowels. Consonants have specific vowel changed via diacritic or mutation.
  • Alphabets: Letters are arranged in a fixed order and represent consonants and vowels. Mix these letters to make other phonemes. The two most well known examples are the Latin and Cyrillic alphabets.
  • Syllabaries: Phonetic with symbols to represent syllables. Examples Japanese hiragana.
  • Semanto-Phonetic: Symbols represent both sounds and meanings. Example kanji.

Linguistics is the analytical study of language.

  • Some sub-fields are syntax (word order), phonetics, semantics, and morphology

Notational Conventions

  • italics : reference to a word, in general
  • // : broad (abstract) notation of speech units
  • [] : narrow (concrete) notation of speech sounds
  • <> : notation of graphemes

Languages have a structure to them Not everything is linear Short vowel indication in most abjads Digraph: Combination of two characters to make a single sound Grapheme: A contrastive symbol, C and K Allographs: Non contrastive R and r Mora: A measure of syllabic "weight"

14 September: Psychology

-Psychology is the study of human thought.

-Margaret Livingstone/ Vision & Arts: The Mona Lisa: Dr. Livingstone noticed that Mona Lisa's smile changes depending on where her eyes were focused on. She explains this elusive smile using the fovea and the peripheral area. The fovea is the area of the eye where human see colours and object's details. On the other hand, the peripheral area does not see colour (black and white) and it also sees shadows and motions. Therefore, when people focus on Mona Lisa's eyes, the peripheral area is seeing the shadow of the cheekbones, making it appears that she is smiling. However, when people focus on Mona Lisa's mouth, the smile is gone.

-Cognitive Psychology: Is the study of human thought or the way in which a brain processes information.

-Santiago Ramon y Cajal: It was suspected that the brain is composed of connected lattices of fine thread. Therefore, Cajal found a way to stain the neurons in the brain which shows the appearance of the entire cell. This proved that the entire cells have different shapes and sizes. Using this technique, Cajal found that the neuron is composed of three parts: the cell body, dendrite, and axon.

-William James: James agreed with Wundt with the importance of immediate experiences and introspection, which is the examination of a person's own emotions). However, he disagrees with the fact that conscious should be broken into two elements: feeling and sensation. Instead, James focused on the function of mental processed and how they help people adapt to the environment. This is called Functionalism.

-Wilhelm Wundt: Wundt believed that consciousness should be separated into two elements: feeling and sensation. Also, he believed that all scientific psychology should only focus on a person's consciousness.

-Edward Titchener and his problem with Introspection: Instead of emphasizing the relationship of feeling and sensation, he believed in identifying the elements themselves and relying less on introspection.

-John Watson: Watson believed that it is hard to study a person's subjective experience. For Watson, science should be replicable everything the same experiment is performed. Therefore, he proposed that scientific psychology should be focusing on the study of behaviors, thing that a person can see, not experience.

-B. F. Skinner: Skinner wanted to know how animals learn when they are in situations like finding food, shelter, and mates. Therefore, he built an experiment called conditioning chamber (Skinner Box). In this experiment, a rat is inside a box and everything it presses the lever, food is given. It was found that the amount of time that the rat presses the level increased over time. This proved Reinforcement, which states that the consequence of behavior determine if a behavior is likely to occur again or not.

- Noam Chomsky: According to Chomsky, a child can generate sentences that they have never heard before, which means that learning a language was a result of reinforcement. He believed that language relies on rules a people use to understand and produce words and sentences. In addition, Chomsky also responded to the book "Verbal Behaviour" by B. F. Skinner. He critiques that some important features of language were ignored when using behaviorism.

-Donald Broadbent: Broadbent is well-known for his work "dichotic listening". In this experiment, a person is asked to listen to two distinct speech simultaneously, one presented to each ear using headphone. Afterward, the subject was asked to determine what he/she heard. It was found that the subject is most likely to recall what was said in each ear rather than mixing both information. Based on this result, Broadbent suggested that a human mind can recieve multiple channels at once.

There are three main approaches to cognitive psychology

  • Experimental cognitive psychology
  • Cognitive neuropsychology
  • Cognitive neuroscience

Aphasias: Woenicke's area where speech is fine but nonsensical and they can't understand questions. Broche's area where it is hard to speak but can understand questions.

19 September: Philosophy

Philosophy: epistemology, metaphysics, and value theory.

Value theory (axiology):

  • Ethics: things like virtue ethics
  • Metaethics: what is ethics
  • Aestethics:

Metaphysics: focuses on questions about fundamental concepts such as being, existence, and reality.

Epistemology: philosophy of knowledge. Focuses on questions such as, what is knowledge, how do we know things, and more.


Turing test: The computer has a conversation with a human by imitating a human. If the computer convinces the human that it is human, it passes the test. Turing explained the test is not about testing intelligence, but if the tester can imagine intelligence in the computer.

Counterarguments to Turing test:

  • Technological limitations: the test is too difficult, the machine necessary for passing the test is too expensive
  • Too easy: not justified by a theory of intelligence, does not test for mind, testers could be gullible

Functions: have inputs and outputs

Turing Computable Functions (Turing Machines): a function is Turing computable if and only if a Turing Machine can compute it. A Turing machine can compute any computable function.

21 September: Computer Science

General Definition of Algorithm - a set of clearly defined steps that will lead to an optimal result 100% of the time

Attributes of Computational Algorithms (Knuth): Knuth's definition of Algorithm

  • Finiteness - an algorithm always terminates, not unlimited
  • Definiteness - every step precisely defined and unambigious
  • Effectiveness - each operation must be primitive enough that a human can perform it
  • Input/Output - an algorithm must have one or more inputs and an output


  • based on common sense, a rule of thumb
  • i.e. "Yes, and" - a heuristic in improv- aim to never negate what the last person said- accept what they have said and expand on that line of thinking
  • used when it would take too long to arrive at the correct or best solution - used day to day

Logical Expressions:

  • and:
  • When two propositions are conjoined by and. The expression is true when both propositions are true.
  • or:
  • When two propositions are conjoined by or . The expression is true if either of the propositions is true.

Module Two: Verbal Ambiguities

26 September: Sources of Ambiguity

Syntax vs. Semantics: Syntax : arrangement of words and phrases to create well-informed sentences in a language, it is the physical shape of a thing ; basic physical properties. while semantics relates to understanding the meaning of word. phrase, sentence or a text, it determines whether sentences in which it occurs are true or false.

Semantics: meaning of a word, sentence, or phrase.

Propositions: things that can be either true or false, such as declarative statements. Propositions can transcend languages, but in doing so, their truth value might change.

Truth functions: functions that work with propositions and output is either true or false. This output is determined by the truth values of the given inputs.

  • AND, OR, NOR are examples of truth functions.
  • BECAUSE is not a truth function.

Ambiguity: A sentence is ambiguous if it can be interpreted with more than one distinct meaning.

  • Lexical: caused by a word that can be understood in more than one way.

Example: The fisherman went to the bank.

  • Structural: when words in a sentence can be grouped differently.

Example: ((I saw) (an (elephant( in (my pyjamas))))). vs (I saw (an elephant) (in (my pyjamas)))

Polysemy: openness to different interpretations, but only one semantic core of the sentence or phrase.

Vagueness: can made from a unmade semantic decision or imprecise boundaries. Example: using words such as "good" and "large" that are relative to people or situations.

Context Sensitivity: quality of words whose meaning can be interpreted differently depending on the context they are placed. Example: The fisherman went to the bank and caught a fish. vs. The fisherman went to the bank to deposit his paycheque.

28 September: Ambiguities in the Perception of Language

Phoneme: perceptually distinct unit of sound that distinguishes a word from another.

  • Phonemes are perceived categorically in the brain
  • Potentially discriminable sounds are assigned to functionally equivalent categories
  • Phoneme categories are formed during critical period of language acquisition.
  • Phoneme categories will vary between individuals and languages

Perceptual magnet effect: when discriminability between vowels is reduced near prototypical sound.

Headturn preference test: test used to determine phoneme preference in children. Children are conditioned to expect a response with one type of phoneme and not the other. Then they are tested with both.

Perceptual landscape: perceptual categories of an individual developed through experience.


  • Context: Stimulus set - stimulus from previous sounds can affect perception (patterns can cause expectations)

Adjacent phonemes- adjacent phonemes can change their pronunciation. So it can help to solve previous ambiguity and also guess future sounds. Acoustic rate of speech - rhythm can have an effect Lexical context - a persons lexicon can affect what their brain perceives.

  • Experience + adaptation

Example: lexical capability

  • Multi-sensory integration:

Visual - Head movement and mouth movement: McGurk effect

Audio- Audio sensory areas activate during lip reading exercises.

Motor - motor areas associated with the lips activate during speech perception and aid in diciphtering audio information, in children imitation is also important for learning to discriminate sound.

3 October: Anaphora

Key Terms from the October Third Lecture:

Parse trees - Trees used to distinguish structural ambiguities in linguistics. Words are grouped together as "branches" on the tree in order to indicate particular interpretations.

anaphoric pronouns - Pronouns that refer to a subject used before the pronoun in the sentence / previous sentence.

cataphoric pronouns - Pronouns that refer to something after themselves in a sentence.

non-anaphoric pronouns- Pronouns not referring to previous information. Example: "We"

  • Anaphoric proper names:
  • Anaphoric definite descriptors: Examples: The Queen has owed more than 30 corgis during her reign. "Her Majesty" is a descriptor

The 42nd President of the United States of America's wife did not win the election. " 42nd President of the United States of America's wife" is a

donkey anaphora- Well formed English sentences very easy for the native speaker to understand but often quite difficult to translate into formal logical language. E.g Every farmer who owns a donkey beats it.

non-pronominal anaphora- Example: Anne took out the garbage. Helen did too. Here the anaphora is caused by "took out the garbage".

Holistic problem- refers to how people use holistic knowledge of the world and their experiences to resolve ambiguity in sentences, which is not possible for machines.

Computation complexity- ?

5 October: Models of Ambiguity Resolution


  1. Search: find possible antecedents
  2. Match: match anaphoric pronoun with possible antecedents and discard any that are impossible based on gender, plurality, and animacy.
  3. And
  4. Solve with
  5. Heuristics: use heuristics to choose from remaining possible antecedents.
  • First mention: pronoun refers to the first mentioned possible antecedent
  • Parallelism: pronoun refers to the word with the same grammatical role in the previous sentence.

Coherence relations:

  • Cause and effect: one event causes the other
  • Resemblance: an analogy is made between the two events. There is a common idea to both events.
  • Contiguity: events are related by time and space(placement/environment).

10 October: Review of Ambiguities

Module Three: Dynamic Processes

17 October: Plasticity and Development

Dynamic Processes

Dynamic Systems Theory -> A way of explaining how systems with many diverse parts interact to produce ordered patterns

  1. In such a system there a variety of possible patterns, states, but only a few are shown / favored. Why? (this is the ? behind DST)
  2. Tend to be self organizing
  3. Causality hard to pin down

DST Key Terms

  • Collective Variables -> The variables that describe the system as a whole (i.e velocity and angle in a pendulum swinging)
  • Space State -> Abstract representation of a space that represents all possible states in system
  • Attractor states -> the pattern a system tends to settle into (Often represented as a basin in a space state diagram, and called an Attractor basin)

Barn owls lol.

19 October: Motor Control

Wittgenstein argued that no process in the brain is correlated with thought. This is contrary to Chomsky's guiding idea that any system that is able to cope with human verbal behaviour must employ transformable representations such as everyone's favorite parse trees from the lectures on ambiguity in order to analyze ambiguity. Wittgenstein and Quine think we needn't have these representations in our head at all, that an order can precede out of chaos (think Dynamic Systems Theory!).

That doesn't fit with our classic big squishy pink computer theory of what is the brain, and there are other key differences as well. Cognitive functioning in the human brain has a "graceful degradation" instead of the characteristic patterns of breakdown in computers that usually involve whole chunks of functionality failing at once. Computers and brains also learn very differently, and computers proceed in step by step systematic ways whereas brains don't.

Traditional computational models also do not resemble the the brain on a basic level. The functional unit of the brain is a neuron which consists of a dendrites that surround a cell body and receive input from other neurons. Extending from the cell body is the axon, along which an electrical signal propagates (travels) and is passed on to other neurons through synapses.

The McCulloch-Pitt Model of a Neuron is utilized in connectionist computation is modeled after these neurons. A McCulloch-Pitt neuronal model takes different input (usually from other McCulloch-Pitt neurons) between one and zero, multiplies them by a weight, sums them and then feeds the result of the sum into an activation function that determines what value is passed on (usually to other McCulloch-Pitt neurons).

Neural networks are computational systems made up of multiple layers of McCulloch-Pitt neurons (typically at least three). At each layer, each neuron passes its output to every neuron on the next level. Neural networks are generally trained through backpropagation of error. A large database or training set of inputs is fed through the network. The output of the network is then compared with the desired output, and the weights of the various neurons adjusted to correct for the error. Such networks have shown surprising aptitude for identifying certain things in pictures.

The information encoded in neural nets during the training process typically has three features. One, the information is distributed meaning that each of the representations generated are composed from several independent parts of the system. The information is overlapping, in that different representations generated are composed from overlapping parts. Finally, these representations are holographic. Any given part of the representation carries information about every part of the content represented by it.

These representations help give an account of how things like grammatical disambiguation and its other generative and interpretive grammatical activities might occur in the human mind without a model that includes an internalized set of structures. A neural network tasked with a grammatical task wouldn't need a representation and perhaps neither do we.

This fits with other trends in modeling the computation around human behaviour. Traditional "good old fashioned AI" approaches to tasks like catching a ball generally give an account that involves a set of differential equations solved to determine where the ball might land. It is much more likely that humans simply run while looking at the ball so that the angle of their gaze increases at a decreasing rate.

These trends in thought can also be categorized as moving away from representations into dynamics. One example given in lecture of such coordination without control was the Governor, which monitored steam engine speed through utilizing the feedback of a centrifugal pendulum that changed with the speed of the operation of the engine to change the amount of speed in the system.

24 October: N-grams and Language Change/UBC WIKI

1. Go to Ngrams site, click embed chart.
2. Only copy the content inside the double quotes src="your_iframe_url_without_quotes"
3. Read the following
4. Paste what your copied into here right after url=, then paste into your wiki edit. You can also try to change width and height
{{#widget:Iframe |url=your_iframe_url_without_quotes |width=1024 |height=768 |border=0 }}


  • use commas to separate phrases
  • * = wildcard search, fills in * with top results
  • _INF = inflection search, shows different inflections of the word
    • eg. "go_INF on a trip” => “going on a trip”, “went on a trip”
    • eg. “feed the dog_INF” => “feed the dog”, “feed the dogs"
  • _* = part-of-speech tag, shows use of word as different parts of speech
    • eg. “slide_*” compares it’s use as a verb or noun
  • *_VERB = unspecified verb
  • *_NOUN = unspecified noun

26 October: Sleep/Wake and Other Cycles in the Brain

This was the last lecture on dynamic systems.

Critical periods: the brain can stabilize after critical period and will have some basic ideas of the world around us.

Energy landscape with attractor basins and attractors, think for example of navigating: a route in your brain that your are more likely to follow.

Hebb’s cell assembly: allows neurons to participate in different groups. The theory is often summarized as "Cells that fire together wire together". → is there another way without physical connecting all the cells? → oscillatory (f.e. alpha band) activity, (first record by Hans Berger (EEG)). Alpha activity is high when people are bored. Oscillations can organize cell activity: → the temporal definition of cell assemblies is therefore different from the strict connectivity-based definition of Hebb. … Through assembly organization, time is translated to neuronal network space.

Brain composed of multiple oscillators: we have a huge range of frequencies. Slower frequencies (oscillations) can manage faster oscillations.

It is possible to actively manipulate your oscillations → This is covered in the reading.

You have activity being generated in individual cells → field potential changes → activity generated in individual cells → DYNAMIC SYSTEM

Waking/REM/non-REM: Slow wave=mostly in NON REM Then in second half you have REM sleep System will compensate if you are trying to cut on it → f.e. sleep paralysis: you fall in REM sleep very early on (no slow wave), so you are awake but also dreaming.

Sleep experiment: slow wave sleep is key in memorizing.

Another study: Stimulation at different frequencies in slow oscillation sleep → enhances memory as well

Another study: Oscillations have to ability to coordinate activity in different cells.

How to combine sleep and memory? → two-stage model of memory consolidation


Gamma oscillations: 30-80 Hz range. Sensory binding problem, think back of: Gestalt psychologists: The attributes of the object are not in the object, you recognize the whole before you recognize the part.

Synchronization activity could help sensory binding.

→ Moonet faces: people

Gamma oscillations correlates with the ability to disambiguate and recognize.

2 November: Group Project Proposals

Module Four: Mediated Communication

7 November: Avatars and Online Communication

How is language used? Talking to yourself to figure out a problem Used in some plays Meta-linguistic representation, using language to talk about language. Face-to-face communication is the best way to learn language.

  • Features of Language: (face to face conversation)
  • Self-expression – participants take actions as themselves. Eg. Calling someone on the phone in the office, they don’t act as themselves, they put on a persona.
  • Simultaneity- they can produce and receive and once, at the same time.
  • Audibility – the participants can hear each other.
  • Visibility – the participants can see each other.
  • Self-determination – participants choose for themselves what actions to take when. People are not self-determinant when in a customer service setting, because they are given some sort of script. Like what to say to customers etc.
  • Recordlessness – the participants’ actions leave no trace or artifact. Eg. Snapchat claims to keep no record of things, which is making things closer to conversations
  • Copresence – the participants share the same physical environment. Not quite the same thing as a virtual reality. E.g.: professor lecturing to an audience. Counterexample: speaking to a friend who lives far away on the phone. Aerotactile ?
  • Instantaneity – they perceive each other’s actions at no perceptible delay. Lags can through off conversations.
  • Extemporaneity – the participants formulate actions on the spot in real time.
  • Evanescence – the medium fades quickly.

What is the basic setting of language? Face-to-face conversation. Audio-visual could be skyping or video messaging, so that is not the basic setting.

Which of the following is not an example of a participant as defined by Clark?

The listener

  • Bystander
  • Eavesdropper

The monitor.

What part of basic setting does the avatar chat with virtual reality violates?

What part of basic setting does the avatar chat virtual reality incorporates?

9 November: Theories of Mental State Attribution

Using a WIKI=

Follow this link to learn how to use a wiki: