Cognitive Neuropsychology
             (c) 2001-11-15 Fabian M. Suchanek
    http://www.mpi-inf.mpg.de/~suchanek/personal/texts/summaries/newsigh.txt


This is a summary for the "Cognitive Neuropsychology" seminar held
at the University of Osnabrueck in the WS 2001 by Franz Schmalhofer
and Thilo Kellermann. It is merely a resume of the reading assignments
in the book "Cognitive Neuroscience - The biology of mind" by
Gazzaniga. I have restructured the information in order to group
similar data and eliminate redundancy. Since the second midterm exam
only compromises topics of the second half of the semester, this file
is divided into two independent parts.

* First midterm exam
 * The history of Cognitive Neuroscience:
   All important and unimportant theories and people
 * Brain anatomy:
   The labelling of brain areas and brain disorders
 * The methods of Cognitive Neuroscience
   Methods of physically analyzing the brain
 * Cognitive Neuropsychology:
   Cognitive experiments with people

* Second midterm exam
 * Memory Systems
 * Language and the Brain
 * Lateralization
 * Executive Functions

By reading the following text, you accept that the author does not
accept any responsibility for the correctness or completeness of
this text. If you have any corrections or remarks, please send
me a mail. This is the only way to make the publication of this summary 
useful for me, too.

My e-mail address is f.m.suchanek@zweb.de, but the 'z' has to be 
removed from the address.

*********************************************************************
  First midterm exam
*********************************************************************


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  The History of Cognitive Neuroscience
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Phrenologist theory (F.J. Gall & J.G.Spourzheim 1810-1819):
 The brain is organized around 35 specific functions, each stored in
 on specific region in the brain. When a person makes
 more use of one function, the appropriate brain area grows.
 Consequently, there will be a bump in the skull above this region (!)
 and the individual's personality can be analyzed by carefully
 examining all the bumps in his skull.
 Evidence: none
 Truth: none

Aggregate field theory (P. Flourens 1824):
 The whole brain participates in behavior, there are no specific
 functions in specific brain regions.
 Evidence: Brain lesions in birds did not impair a specific behavior
 Truth: Basic functions _are_ localized, elaborate functions are
 distributed. They can be accomplished in numerous ways.

Anti-localizationist theory (P. Marie):
 Certain brain areas are not responsible for certain functions.
 Evidence: Not all patients with a lesion in Broca's area show
 language impairment.
 Truth: The critical brain area just shifted from one place to another
 during development, the function is nevertheless localized.

H.Ebbinghaus' contribution (188#):
 Ebbinghaus found that also more internal mental processes like memory
 can be measured.

F.L. Golz's contribution (1881):
 Golz had removed parts of his dog's brain and the dog
 remained functional.

H.v. Helmholtz's contribution (188#):
 Developed an apparatus for measuring the velocity of nerve conduction.

J.H. Jackson's contribution:
 A topological map of the body is represented in a special brain
 region. Seizures in epileptic patients trigger a certain behavior.

J.H. Jackson's contribution 2:
 One has to distinguish localization of a symptom and localization of
 a function. When a certain brain lesion causes a certain symptom,
 it does not follow that this area alone was responsible for only one
 function.

C.v. Manakow's contribution (190#):
 Damage for one part of the brain can cause problems for another
 part (Diaschisis).

H.Head's contribution (1903):
 Head sectioned a branch of his own (!) radial nerve, instead of
 lesioning animals for his studies.

Whole-system theory (H.Head 1903):
 A lesioned brain is a completely new system, not the old one with
 one part missing.
 Truth: none

R.Cajal's contribution (1906):
 Neurons are distict entities, as discovered with Golgi's staining
 method.

K. Brodmann's contribution (1909):
 Brodmann distinguished 52 areas all over the brain by their
 different cellular structure. It was later found out that
 physiological correlates are associated with these anatomical
 differences.

E. Thorndike's contribution (1911):
 A response which is followed by an award is stamped into the organism
 as a habitual response.

J.E. Purkinje's contribution (191#):
 Purkinje described the first nerve cell.

S. Freud's contribution (191#):
 Freud came up with the idea of distict neurons after his work with
 a crayfish.

C. Sherrington's contribution (19##):
 Sherrington coined the term "synapse".

I.Hyde's contribution (1921):
 Hyde developed a microelectrode and was the first woman to be
 elected to the American Physiological Society.

C. Woolsey & P.Bard's contribution (1930):
 Each modality sends data to more than one sensory map in the brain.

Nativist theory (??? 19##):
 Most aspects of personality are built-in from birth.

Associationist/Behaviorist/Learning theory
 (J.B. Watson, B.F. Skinner 19##):
 Everybody has the same neural equipment and only learning forms
 personality and abilities. There are no built-in abilities.
 Truth: Partially true, some functions are built-in.

Innate grammar theory (Chomsky 1950):
 Since the associationist theory cannot explain how we learn language,
 the ability to acquire a language must be built-in in form of an
 innate grammar.

Gestalt theory (??? 19##):
 Gestalt psychologists worked with perceptual phenomena and found out
 that e.g. apparent motion is built-in and does not have to be learned.

Cell assembly theory (D. Hebb 19##):
 Any set of neurons can learn anything.

G.A. Miller's contribution (1950):
 Cognition must be analyzed instead of behavior.

D. Marr's contribution (19##):
 The brain computes. Neural computation can be understood at multiple
 levels of analysis.


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  Brain anatomy
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

See also: neurobio.txt

Brain damage:
 Brain damage can occur from vascular problems, tumors,
 degenerative disorders, progressive disorders, strokes and trauma.
 Brain damage is often diffuse, rendering it difficult to associate
 a behavioral deficit with a structure.

Vascular problems:
 Caused by an insufficient supply of oxygen and glucose to the brain.

Tumors:
 A tumor is a mass of tissue that grows abnormally. "Benign" tumors
 do not re-occur after removal, while "malignant" ones do. The
 malignant tissue can invade the bloodstream and thus be carried to
 the brain.

Progressive neurological disorders:
 Destruction of brain tissue, often caused by a virus (such as e.g.
 HIV).

Degenerative neurological disorders:
 Diseases such as Alzheimer's, Parkinson's  and Huntingtons'.

Stroke:
 When a foreign substance blocks the stream of blood, the supply of
 the brain is suddenly interrupted, causing a stroke.

Trauma:
 A trauma is a head injury. One distinguishes open and closed head
 injuries, depending on the state of the skull after the incident.

Epilepsy:
 Epilepsy is an abnormally patterned activity in the brain.

Multiple Sklerosis (MS):
 MS is a disease which demyelates the nerve fibers.

Gray & White Matter:
 There is gray and white matter in the brain. The gray matter
 surrounds the white matter, whose color originates from the
 myelin sheets of the nerve fibers.

Connections:
 Axons can connect widely distributed regions. They are then called
 <source>-<destination> connections (e.g. "thalamo-cortical
 connection").

Topographic maps (homunculi):
 There is one sensory map of the body and one motorical map of the
 body in the cortex. These maps represent each region of the body
 with a size according to the regions's sensitivity. Such a
 correspondence between brain area and function is seen only in
 the primary sensory and motor areas.
 Other map-like structures can be found in the visual cortex and in
 the auditory cortex.

Broca's area:
 Location: left frontal lobe
 Tasks: Speaking

Wernicke's area:
 Location: posterior left hemisphere
 Tasks: Comprehension

Brain lobes:
 Frontal (front), parietal (above), occipetal (back), temporal (side).

Primary cortices:
 * primary olfactory cortex: above the nose
 * primary visual cortex: in the back of the head
 * primary motor cortex: in the middle of the parietal lobe, reaching
   from left to right
 * primary somatosensory cortex: parallel to the primary motor cortex

Planes:
 * horizontal plane
 * coronal plane: vertical plane reaching from left to right
 * sagittal plane: vertical plane reaching from front to back


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  The Methods of Cognitive Neuroscience
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

(This is everything which has to do with physically analyzing the
 brain, for experiments, see the chapter on Cognitive Neuropsychology)

Neuroanatomy:
 The study of the nervous system's structure.

Neurophysiology:
 The study of the functions of brain areas.

Neurology:
 The study of brain-damaged persons.

Neurosurgery:
 Is the discipline of cutting away certain parts of the brain in
 order to reduce seizures.

Histology:
 The study of tissue structure through dissection.

Cell morphology:
 The study of how synapses form. The genetic code constraits the
 target region of a connection, but its exact form is shaped
 by experience. The arborization is most complex in early development.
 Later, many braches are eliminated and the ones remaining
 become thicker.

Staining (C. Golgi 1906):
 A specific neuron is impregnated with silver. Thus, the whole cell
 with all its dendrites and axons can be visualized. However, only 1%
 of the cells will absorb the stain.

Retrograde-tracing:
 A retrotracer (e.g. HRP, horseradish peroxidase) is injected into
 the axon and is transported back to the soma. Then, the
 animal/human is killed and the axonal connection can be seen.
 Antrograde tracers, in contrast, are injected at the dendrites and
 visualize the connection to the soma.

Electrical stimulation:
 Electrical current is lead into a specific area of the brain,
 possibly triggering a certain action or thought. Firstly done
 systematically by W. Penfield in the 1940s.

Single cell recording:
 An electrode is inserted into the brain and measures the electrical
 potential as caused by the surrounding neurons. Computer algorithms
 then differentiate the pooled activity and calculate an individual
 neuron's contribution. Single cell recording could also be done
 intracellularly (and would thus really record a single cell), but
 inserting the electrode often destroys the cell.
 In spite of the advantages of this technique, the brain can be better
 understood by analyzing fire patterns of groups of neurons rather than
 single cell activities.

The problem of cause and effect:
 Single cell recording is a correlational approach, it can hardly be
 determined which neuron really causes a response. Reciprocal
 (circular) connections make it difficult to determine the cause of
 a brain activity. Nevertheless, latency measurements and time
 comparing can yield hints about which activity preceded another one.

Lesioning:
 Up to now, scientists only lesion animals' brains for
 their studies. For studies on humans, patients with naturally
 occuring lesions are still needed. When lesioning, it is difficult to
 selectively destroy a certain brain area. Nowadays, lesions can also
 be produced by chemicals (like MPMP). Even temporary lesions can be
 created by pharmacological manipulations or by injecting chemicals
 which freeze the brain area. These lesions go away when the chemical
 fades off.

Computer tomography (CT):
 X-rays (Roentgenstrahlen) are sent through a living person's brain
 in 3 dimensions. Since the density of tissue affects the absorption
 of x.-rays, a 3d map of the brain can be calculated.
 Spatial resolution: 1/ 5-10 mm

Magnetic resonance imaging (MRI):
 Hydrogen atoms in the brain are dipoles. They are normally randomly
 oriented, but can made turn to one direction by an external magnetic
 field. Radiowaves are then passed through the brain and make the
 atoms spin. When the radiowaves are turned off, the atoms return
 to the orientation of the external magnetic field. By doing so,
 they produce small local magnetic fields which can be measured.
 Spatial resolution: 1/ 0.1 mm

Functional magnetic resonance imaging (fMRI):
 fMRI works like MRI, but concentrates on hemoglobin, the oxygen
 carrier of the blood. Oxygenated and deoxygenated hemogloboin can
 be distinguished by fMRI, their ratio is determined and identifies
 the most oxygen-demanding (= most active) regions in the brain.
 Their MRI signal is called "BOLD-signal".
 Spatial resolution: 1/ 0.1 mm
 Temoral resolution: 1/ 6 seconds (low)

Subtraction method:
 A fMRI is done under experimental and under control condition. The
 difference of the two BOLD-signals is calculated and thought to
 identify the brain region involved in the experiment.

Event-related fMRI:
 Works similar to ERPs, but is independent from a control condition.
 It is not only possible to identify involved brain areas but also
 to track the dynamics of neural activity.

Electroencephalography (EEG):
 If large groups of neurons are active together, they produce
 electrical potentials large enough to be measured by placing
 electrodes on the scalp. Certain behavioral states can be associated
 to different EEG signatures.
 Spatial resolution: very low
 Temoral resolution: 1/ 1 millisecond (very high)

Event-related potentials (ERPs):
 When brain activity is measured in response to a particular task,
 the resulting EEG minimally differs from an EEG in a control
 condition. This difference is the Event-Related Potential. Since the
 ERP is only a tiny signal embedded in the ongoing EEG, many trials
 are needed in order to calculate a significant average ERP.
 ERPS can hardly elucidate the function of specific brain areas but
 they can tell us something about the time course of processing in
 the brain.

Magnetoencephalography (MEG):
 MEG measures the local magnetical fields of active neurons. Although
 MEG has a good temporal resolution and can also identify the location
 of activity, it can only measure neurons parallel to the surface
 of the skull. With MEG, event-related fields (ERFs) can be measured.

Positron emission tomography (PET):
 A PET scanner is a gamma ray detector: A tracer (an isotope) is
 injected in the blood and gets distributed in the body and the brain.
 Whenever an isotope decays, it emmits a positron. If this positron
 collides with an electron, two photons are emitted in contrary
 directions. The PET scanner registers them and calculates the
 location of their origin. PET scanners thus locate amounts of blood
 and hence the most active regions of the brain.
 Spatial resolution: 1/ 5-10 mm
 Temporal resolution: low

fMRI compared to PET:
 + fMRI is "noninvasive" (i.e. needs no tracer injected into the blood)
 + fMRI studies can be "longitudinal" (i.e. carried out over a long
   period of time)
 + fMRI has a higher spatial and temporal resolution
 + fMRI sessions can do functional and anatomical recordings at the
   same time

Transcortical magnetic stimulation (TMS):
 The activity of a brain region can be altered by applying an
 external magnetical field to the head.

Angiography:
 A dye is injected into the blood and then the patient undergoes an
 x-ray study. Thus, the major arteries and veins and the distribution
 of blood can be seen.

Fourier-Transformation:
 ...means analyzing a wave and identifying its component waves.


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  Cognitive Neuropsychology
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

See also: cogpsy.txt

Key concepts of CogNeuPsy:
* Information processing depends on internal representations
* Mental representations undergo transformations

Mental operations/transformations:
 Mental Op's take a representation as an input, perfom some process
 on it and then produce a new representation as an output.

Letter matching experiment (Posner 1986):
 Task: Two letters are presented and the subject has to decide whether
 they belong to the same category (vowels or consonants).
 Result: Subjects respond fastest when the letters are the same.
 Conclusion: We derive multiple representations from stimuli, first
 physical ones, then phonetic ones and last category ones.

Letter matching experiment 2 (Posner 1986):
 Task: Two letters are presented with a short time interval in between
 and the subject has to decide whether they belong to the same
 category (vowels or consonants).
 Result: Subjects needed less time to answer.
 Conclusion: During the time interval, the 1st stimulus is already
 transformed to an internal, more abstract code. Thus, the subject
 only needs to transform the second one, saving the time.

Membercheck experiment (Sternberg 1957):
 Task: A set of numbers is presented and afterwards, the subject has
 to tell whether another number was in this set.
 Result: Reaction time increases with set size.
 Conclusion: The memory comparison operation takes a fixed amount of
 internal processing time per item, it appears to be serial and
 exhaustive. (Which is astonishing: When you run through an array of
 n numbers and look for x, you can stop if you found x and do not need
 to run to the end of the array. This method thus takes _less_ than n
 steps. To my view, this suggests that the notion of a serial array
 is not adequate here.)

Word superiority effect:
 Subjects are most accurate when stimuli are words.

Dual tasks:
 For dual task studies, performance on a primary task alone is compared
 to performance on that tasks concurrently with a secondary task. Thus,
 interference of mental operations can be shown. Experience and
 learning can improve performance on dual tasks.

Resource conflict:
 If two tasks require the same resource (e.g. the verbal system),
 simultaneous performance on these tasks cannot be improved by
 learning.

Computer modeling:
 The computer is given input and then must perform internal operations
 according to a model to create behavior.
 + Successes and failures of a model give valuable insights to a
   theory's strengths and weaknesses.
 + Since computer models must be totally specified, the researcher has
   to develop a completely explicit theory.
 + Models can generate novel predictions which can be tested with
   real brains.
 - Computer models are always radical simplifications
 - Computer models are often not biologically adequate
 - Modeling efforts are restricted to a relatively narrow problem

Symbolic and connectionist models:
 Symbolic computer models include units that represent symbolic
 entities. In contrast, processing (of e.g. symbols) is distributed
 over the whole system in connectionist models. Connectionist models
 can be used to demonstrate how a systems behavior changes after
 lesioning its cognitive system.

Vehicle experiment (Braitenberg 1984):
 Task: Two vehicles are built, each with two photosensors at their
 front and 4 weels. In one vehicle, the left sensor is connected to the
 engine of the left weel and the right sensor to the right weel's one.
 In the other vehicle, the two wires cross.
 Result: Astonishingly, the first vehicle heads away from light while
 the second pursues it. This seems to have something to do with the
 wires!
 Conclusion: When the connection of the wires (=neurons) is different,
 different behavior results!

Single dissociation:
 When two tasks just differ in requiring one hypothetical mental
 operation, the comparison of performance on these tasks can be used
 to identify this operation. One speaks of single dissociation if
 an impaired and a health ycontrol group are tested on these two
 tasks and a between-group difference is apparent in only one task.

Double dissociation:
 Two minimally differing tasks are designed and given to two groups,
 each with a special impairment. If one group failes on the first
 task, while the other group failes on the second task, one speaks
 of double dissociation. Double dissociation reliantly identifies
 a mental operation.
 Double dissociation tasks can also be carried out with healthy
 subjects: Evidence of separable cognitive operations can be gained
 by demonstrating that one task is affected by one type of
 manipultion whereas a second task is selectively affected by a
 different manipulation.

Group studies:
 Group studies have been criticized as inappropriate for human
 neuropsychology because of the variability among patients
 assigned to the same group.

Reaction time experiments:
 By measuring the latency of a subject's answer, we can conclude about
 the accessibility of information in the brain. The latency can either
 be measured in general (as in the Stroop task, the letter matching
 experiment or the word superiority effect), or differentially by
 giving the subject a prime and thus decreasing the latency in a
 subsequent task (e.g. word stem completion experiment).


*********************************************************************
  Second midterm exam
*********************************************************************


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  Memory systems
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

see also: cogpsy.txt

Memory: The product of a learning procedure.

Learning: An experience-based modification of behavior.

Modality: Sensory organ.

Levels of processing model theory (Craik & Lockhard 1972):
 There are different levels of processing (superficial,
 intermediate and deep) and deep processing leads to a
 better learning.
 Evidence: When people are to make a judgment about the
 meaning of a word (deep processing), they can later
 retrieve it more easily than when they had to decide
 whether the word was written in uppercase (superficial
 processing).
 Counter-evidence: This theory only works for explicit
 memory.

Modal model theory (Atkinson 1968):
 Memory consists of
 * sensory memory / sensory register
   Sensory inputs involuntarily enter the sensory memory
 * short term memory, STM
   Attention then contributes to the storing in STM
 * long term memory, LTM
   Rehearsal then contributes to the storing in LTM
 The retention time increases from sensory memory to LTM.

Working memory: Short term memory plus sensory memory.

Three-part working memory model (Baddeley 1974):
 The working memory consists of
 * the phonological loop (a temporary audio buffer)
    which is subdivided into
    * a phonological store (speech perception) and
    * an articulatory process (speech production)
 * the visual-spatial sketchpad (a tmp video buffer)
 * the central executive
 Evidence: When people are to maintain auditory
 information, they can be distracted by other auditory
 information, but not by visual information (and vice
 versa).
 Furthermore, each of the above system can be damaged
 selectively by brain lesions.

Long term memory component theory (Tulving 1995, Squire 1995):
 Long term memory can be subdivided into
 * declarative/explicit memory
   * episodic memory (Medial temporal lobe, diencephalon)
   * semantic memory (Medial temporal lobe, diencephalon)
 * nondeclarative/implicit memory (motor cortex plays a role)
   * skills and habits (Striatum)
   * priming (Neocortex)
   * simple classical conditioning
     * emotional responses (amygdala)
     * skeletal musculature (cerebellum)
   * nonassociative learning (reflex pathways)
 Evidence: The patients H.M. and M.S. showed a double
 dissociation for explicit and implicit memory.

Explicit memory: Every memory which can be verbalized.
 * Acquisition: One-trial learning is possible
 * Forgetting: Memory fades with time.
 * Flexibility: Memory is modality-independent
 * Reliability: Contents may be forgotten
 * Consciousness: conscious (noetic)
 * Development: Late stage in development

Implicit memory: Every memory which cannot be verbalized.
 * Acquisition: procedural learning by repetition
 * Forgetting: Memory does not fade with time.
 * Flexibility: Memory is only accessible via learning modality
 * Reliability: Contents are not forgotten
 * Consciousness: unconscious (anoetic)
 * Development: Early stage in development

Episodic memory: Memory for events in one's life.
 * Acquisition: One-trial learning
 * Personal Involvement: Yes
 * Content: Whole event plus context
 * Retrieval: Effortful search in personal past
 * Orientation: Past
 * Development: Late stage in development

Semantic memory: Memory for facts and world knowledge.
 * Acquisition: Multiple trial learning
 * Personal Involvement: No
 * Content: Pure world knowledge
 * Retrieval: Effortless
 * Orientation: Present and future
 * Development: Early stage in development

Priming: The improvement in identifying or processing a stimulus
 as the result of its having been previously processed.

Example for an explicit memory test: Subjects have to learn a list
 of words and are afterwards asked whether a word had been on the
 list.

Example for an implicit memory test: Subjects have to learn a list
 of words and afterwards do a word fragment completion task.

Word fragment completion task: Subjects are presented a string of
 letters with blanks and have to fill the blanks in order to form
 a word.

Serial reaction time experiment:
 Task: Subjects have to place 4 fingers on 4
 buttons. Then 4 lights flash in a repeated sequence and the
 subjects have to press the appropriate buttons.
 Result: Although Subjects do not know that the sequence is repeated
 they become better at pressing the buttons.
 Conclusion: They learn the sequence implicitely.

Amnesia: Memory problem. Usually just concerns explicit memory
 and leaves intact the implicit memory. Often, amestics are
 nevertheless able to learn semantic information at a very low rate.

Anterograde amnesia: Inability to acquire new information, results
 from damage to the hippocampus.

Retrograde amnesia: Inability to recall old information, results
 from damage of the anterior temporal lobe.

Isolated retrograde amnesia: Retrograde amnesia without anterograde
 amnesia.

Source amnesia: Inability to retrieve the episode of one's life
 where something was learned.

H.M.: Well-known neurological patient whose temporal lobes were
 removed bilaterally. As a consequence, H.M. developed a severe
 anterograde amnesia and also a retrograde amnesia extending
 backward from his surgery for 3 years. The transfer from STM to
 LTM was disrupted.

R.B.: Neurological patient who had a lesion in the CA1 pyradimal
 cells in the hippocampus. He also had a anterograde amnesia,
 indicating the hippocampus' role in forming new memories.

M.S.: Neurological patient with a lesion in the right occipital
 lobe. He showed an impairment on implicit memory tasks but
 performed normally on explicit memory tasks. Thus, H.M. and M.S.
 complete a double dissociation for implicit and explicit memory.

HERA, Hemispheric encoding retrieval assymetry: Encoding takes
 place in the left prefrontal cortex. Retrieval of semantic
 information also activates the left prefrontal cortex while
 retrieval of episodic information activates the right prefrontal
 cortex. (*R*etrieval of *E*pisodic information is *RE*chts,
 alles andere links :-)

Consolidation: A subprocess of learning which takes place in the
 neocortex. Consolidation strengthens the associations between
 multiple stimulus inputs and activations of previously stored
 information. It is coordinated by the hippocampus.

Hebb's law (1949): If a synapse is active while a postsynaptic
 neuron is active, the synapse will be strengthened.

Long term potentiation, LTP: LTP means that neural stimulation
 leads to greater synaptic strength, thus provoking larger
 excitatory postsynaptic potentials on subsequent stimuli. All
 in all, weak stimuli are potentiated when they co-occur with
 stronger inputs (cooperativity, associativity, specificity).
 NMDA-receptors might be responsible for producing the LTP,
 since they are doubly gated (see neurobio.txt).

Long term depression, LTD: LTD means the reduction of synaptic
 strength as a reaction to persisting stimuli in a slow rate
 (habitation).


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  Language and the Brain
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Anomia: The inability to name objects.

Aphasia: Deficit in language production or comprehension. Aphasia
 is thought to be a processing impairment rather than a loss of
 knowledge.

Categorizing aphasics:
 Aphasics can be categorized by 3 parameters
 * Spontaneous speech
 * Auditory comprehension
 * Verbal repetition
 (but the book does not make use of this)

Wernicke's aphasia: The inability to comprehend language meaning.
 * Spontaneous speech: Fluent speech, but without meaning
 * Auditory comprehension: Severe difficulties in understanding
 * Verbal repetition: (?)

Broca's aphasia: The inability to produce grammatically correct speech.
 * Spontaneous speech: Effortful, in telegram style
 * Auditory comprehension: No ability to understand grammar
   ("Peter kicks Bob" vs. "Peter was kicked by Bob")
 * Verbal repetition: Often difficult
 (*Bro*ca = *Bro*duction :-)

Agrammatic aphasia: The inability to produce grammatically correct
 sentences. Agrammatic aphasics only say small sequences of 2 or 3
 content words.

Conduction aphasia: Aphasia resulting from a damage to the
 connection between Broca's area and Wernicke's area
 * Spontaneous speech: Problematic
 * Auditory comprehension: OK (?)
 * Verbal repetition: Problematic

Dyslexia: The inability to read.

Progressive semantic dementia: Difficulty to assign objects to
 categories.

Naming difficulties: Patient with different lesions have different
 problems in naming objects. The temporal pole is correlated with
 problems in naming persons, the anterior part of the left temporal
 lobe is correlated with problems in naming animals and the
 posterolateral part of the left inferior temporal lobe is
 correlated with problems in naming tools.

Lexical decision experiment:
 Task: Subjects have to decide whether a string is a word. Before
 every string, a prime word is shown. The time needed to figure
 out that a string is a correct word is measured.
 Result: Subjects detect correct words more quickly when the prime is
 semantically related to the word.
 Conclusion: Priming can result from a spread of activation in a
 semantic net. Activation is set to a specific concept by the prime
 word and has to reach the second word before a decision can be
 made.

Lexical processing:
 * Lexical access (retrieving the word input)
 * Lexical selection (selecting a lexical entry for the input)
 * Lexical integration (combining the current word with the
   words already encountered in order to form a sentence)

Syntactic analysis: Finding a grammatical structure for a
 sentence. Syntactic analysis even goes on in the absence of
 meaning.

Mental lexicon: A store of information on words in the mind.
 It includes
 * semantic information
 * syntactic information
 * word form information
 Words may be added or forgotten. Words used more frequently are
 accessed more quickly.
 A normal adult speaker has a passive knowledge of about 50000 words.
 (see also lingu.txt)

Levels of representation of word knowledge:
 * semantic information
 * lexical information
 * phonological information

Pandemonium model theory (Selfride 1959):
 A model for letter recognition based on "demons", i.e. little
 units which work by themselves.
 * An image demon receives sensory input (pixels)
 * Feature demons decode specific features, each demon gets
   active according to one specific feature (bars, curves etc.)
 * Cognitive demons get activated when a certain combination
   of feature demons is active (letters)
 * The decision demon selects the most active (i.e. most
   plausible) letter

Cohort model theory (Wilson 1980): Lexical selection takes place
 incrementally. It works like scrolling in a list by typing the
 first letters of an expression.

Garden path theory (Franzier 1987): A sentence is analyzed
 according to two principles:
 * principle of minimal attachment: Chose the syntactic
   interpretation which needs minimum number of phrase marker
   nodes
 * principle of late closure: Attach new words to the phrase
   which is currently being processed
 (see also foc.txt)

Macroplanning and Microplanning theory (Levelt): Speech
 conception consists of two phases
 * Macroplanning (planning the idea we want to express)
 * Microplanning (word choice and structure choice)

Image to word theory (Levelt): When a subject is to name
 an object in an image, three levels are passed:
 * On the conceptual level, all concepts related to
   the image are activated
 * On the lemma level, nodes with the appropriate
   syntactic information become activated
 * On the lexeme/sound level, the appropriate sound is
   assembled

Wernicke-Lichtheim-Geschwind model theory (W,L,G 1967):
 Language processing involves 3 brain areas, which are all
 interconnected.
 * "M" (Broca's area, speech planning and grammar, leads to
   *M*otor area)
 * "A" (Wericke's area, phonological lexicon, receives from
   *A*uditory system)
 * "B" (storage of concepts, *B*loede *B*uchstabenwahl)
 Auditory inputs come to Wernicke's area. Here, words are
 accessed. Then the flow of information goes to Broca's area
 where the syntactical structure is analyzed. Both Broca's
 area and Wenicke's area then project to the concept area,
 which causes comprehension.
 Evidence: Lesions of those brain pathways associated with
 the pathways in the model cause impairments as predicted
 by the model.

N400: The name of a brain wave related to linguistic processes.
 Its amplitude is increased when sentences end in anomalous
 words. N400 effects are modality independent.


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  Lateralization
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Corpus callosum: The largest nerve fiber in the brain which connects
 the two cerebral cortices. The splenial (i.e. posterior) area of
 the corpus callosum transports viual information, while the anterior
 part transfers higher semantic information about stimuli.

Homotopic: Property of a brain nerve which points to the same region
 in the other cerebral cortex. Most fibers in the corpus callosum are
 homotopic.

Heterotopic: Property of a brain nerve which points to another region
 in the other cerebral cortex. If a fiber of the corpus callosum is
 heterotopic, i.e. points from region A in the first hemisphere to
 region B in the second hemisphere, then there are usually also fibers
 pointing from A in the first to B in the first.

Ipsilateral: On the same side.

Split-brain patient: A patient whose corpus callosum was cut (in
 order to experiment with lateralization or to eliminate seizures).
 Thus the flow of information between both hemispheres is blocked.
 Split-brain patients are not able to
 * name objects presented to the right hemisphere because language is
   in the left hemisphere. Nevertheless, they can point on the image of
   that object with the left hand (controlled by the right hemisphere).
 * arrange blocks with the right hand (controlled by the left
   hemisphere) because the right hemisphere usually coordinates these
   movements.
 * arrange blocks with both hands because the two hands are in
   competition
 * transfer knowledge teached to one hemisphere to the other
 * find with the right hand an object held in the left hand

What is not disconnected in split-brain patients:
 * crude spatial information can be transferred from one hemisphere to
   the other
 * The two hemispheres rely on a single attention system

W.J.: The first split-brain patient.

Cerebral specialization, lateralization: The fact that the two
 hemispheres specialized in different functions. Cognitive functions
 are never wholly lateralized (except for grammar) and can also be
 learned with the usually non-specialized hemisphere. The two
 hemispheres work in concert to perform a task, but their
 contributions vary. The hemispheres differ slightly on a
 neuroanatomical level, in their strategies and in their
 efficiency on specific tasks. The visual system is more
 strictly lateralized than the other sensory systems.

Competitive vs. cooperative: As found out by artificial neural
 networks, a high corpus callosum connection speed leads to
 cooperating hemispheres. When the corpus callosum is cut or
 one hemisphere is destroyed, the tasks can no longer be
 accomplished. A low speed connection leads to independently
 working hemispheres. Destruction of one hemisphere or the
 corpus callosum does not entail complete impairment on the
 task. The brain has a low speed corpus callosum connection,
 caused by the slow firing rate of neurons and the time needed
 for axonal signal transduction.

Methods of assessing lateralization:
 * presenting stimuli to only one hemisphere (dichotic listening
   experiment or flashing images on one side of the visual
   fixation point)
 * investigation of split-brain patients
 * investigation of patients with unilateral lesions

Dichotic listening experiment:
 Task: Subjects listen to two different sequences of words (or to
 music) which each are presented to one ear at the same time.
 Result: Subjects are better at recalling the words from the right
 ear, but better at recalling a melody from the left ear.
 Conclusion: The left hemisphere is better at language, the right
 is better at music.

Fourier theorem: Any complex pattern can be described as a
 composite of sine functions which vary in amplitude and phase.

Frequency lateralization theory (Sergent 1985):
 The right hemisphere is biased to represent low-frequency data
 while the left one represents high-frequency data. Whether a
 frequency is high or low is decided in relation to the context
 in which the data is presented. The term "frequency" is here
 applied to
 * the size of an object (small -> high frequency)
 * the narrowness of a pattern (narrow slashes -> high frequency)
 * the hierarchical position of subimages within an image
   (local -> high frequency)
 * the pitch of tones (high -> high frequency)
 * the resolution of a picture (high -> high frequency)
 * the content of speech (prosodic data -> low frequency,
   words -> high frequency)
 Evidence:
 * Narrow patterns are judged more quickly with the left
   hemisphere
 * When two sequences of tones of varying pitch are presented to
   the left and to the right ear, subjects report to have heard all
   lower-pitched tones in the left ear and all higher-pitched tones
   with the right ear (the "scale illusion").
 * lesions to the right hemisphere disrupt prosody much more than do
   lesions to the left hemisphere

Representation of spatial data: Spacial data can be represented in
 two ways:
 * categorical (i.e. by spatial relations of objects)
 * coordinate (i.e. by exact positions of objects in space)

Spatial representation lateralization theory (Kosslyn):
 The left hemisphere forms categorical spatial representations
 while the right one forms coordinate spatial representations.
 Evidence:
 * Judging whether two objects are near or far are done more quickly
   with the right hemisphere (metrical judgement)
 * Categorical judgements were done more quickly with the left
   hemisphere
 * Patients with left-hemisphere impairment have difficulties in
   making categorical spatial decisions (e.g. whether two objects
   in one image occur with their positions switched in the second
   image). Patients with right-hemisphere impairment have difficulties
   in making coordinate spatial decisions (e.g. about the change
   in distance of two objects)

Snowball theory (Kosslyn):
 The seed for the coordinate-categorical distinction was language
 which favored categorical representations.

Categorizing lateralization theory (Marsolek 1995):
 The left hemisphere uses prototypes to store information while
 the right one uses an exemplar-based technique.
 Evidence: Subjects were quicker at classifying known objects
 with the right hemisphere, but better at classifying previously
 unseen objects with the left hemisphere.
 In word priming, a change in case (lower or upper case writing)
 reduces the priming with the right hemisphere.

GAD-theory (Corballis 19xx):
 There is a "Generative assembling device" ("GAD"), which generates
 complex representations from smaller units. Since the GAD is in the
 left hemisphere, this would explain the predominance of the right
 hand. Since language was developed initially from the need to
 describe tool use (done with the right hand) language occurs to be
 also specialized in the left hemisphere.

Emotions-are-right theory (??? ???):
 Emotions are lateralized to the right hemisphere.
 Evidence: When images of facial expressions were assembled from
 two left-halves, they were judged more emotional than others which
 were assembled from two right halves.

Valence theory (Damasio 19xx):
 The left hemisphere is more involved in positive emotions (approach)
 while the right hemisphere is more involved in negative ones
 (withdrawl). Approach and withdrawl are very basic behavior
 patterns, which can be observed with nearly any creature.
 Evidence:
 * Patients with left hemisphere damage often suffer from depression
   while right hemisphere damage often leads to mania
 * Watching and anticipating positive stimuli results in an increased
   left hemisphere activity and vice versa.
 * Depressive patients have a higher right hemisphere activity
 * More left hemisphere EEG activity in 3-year-olds suggests a more
   explorative behavior and vice versa.
  (Who helps us memorizing this are our neighbors in the east: The
   Pole will tell us that the POsitive emotions are LEft)

Left hemisphere responsibilities:
 * Language
 * Lexicon
 * Syntax
 * Inferences
 * Generating voluntary facial expressions
 * Categorical space representations
 * Positive emotions
 * Uses prototypes as a representational format
 In general:
 * sequential
 * analytic
 * local
 * high frequency data

Right hemisphere responibilities:
 * Visuo-spatial orientation
 * Can judge grammatical correctness but not syntax
 * Recognizing unfamiliar faces
 * Generating spontaneous facial expressions
 * Music
 * Prosody
 * Coordinate space representations
 * Negative emotions
 * Uses an exemplar-based representational format
 In general:
 * parallel
 * holistic
 * global
 * low frequency data

Gender differences in hemispheric specialization:
 Females:
 * acquire language earlier than males
 * perform better on language tests
 * could be left-hemisphere dominant (but: females have fewer aphasic
   disorders after damage to the left hemisphere)
 * have maybe a lower assymmetry between the two hemispheres (because
   lesions to the left hemisphere cause problems for both verbal and
   perfomance tests)
 Males:
 * perfom better on visuospatial reasoning
 * score higher in maths tests
 * have maybe a higher assymmetry between the two hemispheres

Handedness and left hemisphere language dominance:
 95% of right-handers but only 50% oft the left-handers have a left
 hemisphere language dominance.

Hemispheric specialization in non-humans: Also non-humans have
 specialized hemispheres, especially bird which do not have a corpus
 callosum.


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  Executive Functions
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Iron bolt experiment (Gage 1848):
 Task: The experimenter puts dynamite into a hole in a rock, puts
 an iron bolt atop and lights the dynamite.
 Result: The iron bolt is blown through the experimenter's prefrontal
 cortex. The experimenter survives but becomes impatient and rude.
 Conclusion: The prefrontal cortex is involved in an individual's
 personality.

Subdivisions of the frontal lobe:
 * Motor cortex (most posterior)
 * Secondary motor zones (more anterior)
   * lateral premotor cortex
   * Broca's area
   * posterior portion of the cingulate cortex
 * Prefrontal cortex (most anterior)
   * Lateral prefrontal cortex
   * Ventromedial zone
   * Anterior cingulate
   It is connected almost all cortical and subcortical areas.

Delayed response experiment / Working memory experiment:
 Task: A monkey sees how food is placed in one of two food dwells.
 Both dwells are then covered. A curtain is lowered for a delay
 period.
 Result: When the curtain is put away, the monkey remembers
 which dwell contains the food. If the monkey has (by chance)
 lesions in the lateral prefrontal cortex, it will not remember the
 dwell. Cell recordings showed that special cells are only active
 during the delay period, they "memorize" the stimulus.
 Conclusion: The lateral prefrontal cortex might be responsible for
 working memory.

Associative memory experiment:
 Task: The monkey still sits in the cage and the experimenter puts
 food in one of two dwells. Both dwells are then covered with
 differently marked sheets. A curtain is lowered for a delay period
 and both dwells are exchanged together with their content and
 cover.
 Result: When the curtain is put away, the monkey correctly
 identifies the dwell with the food by the cover. Prefrontal
 lesions do not disrupt this task.
 Conclusion: Associative memory does not seem to be stored in the
 prefrontal cortex.

Piaget's Object permanence experiment:
 Task: A child watches the experimenter hide a reward (for instance
 tasty tasty chocolate chunks). After a delay period, the child is
 encouraged to search the reward.
 Result: Children younger than 1 year cannot accomplish this task.
 Their frontal lobes have not yet matured.
 Conclusion: The frontal lobes prevent a "out of sight ,out of mind"
 mechanism.

Wisconsin card sorting experiment:
 Task: The subject has to distribute cards, which show various
 symbols, from a stack into 4 slots, which are also marked by various
 symbols. The experimenter tells the subject after each card whether
 the placement was correct according to special rules. As soon as the
 subject got the rule and did 10 correct placements, the experimenter
 changes the rule without telling the subject.
 Result: Patients with damage to the lateral prefrontal cortex tend
 to apply the same rules over and over again although the
 experimenter has changed them.
 Conclusion: The lateral prefrontal cortex seems to serve a certain
 mental flexibility.

Abstract shapes experiment (McCarthy 1994):
 Task: Subjects watch shapes occurring in different places on a
 monitor. As soon as one shape occurrs in a place which had already
 previously been occupied, they have to raise their finger. In the
 control condition, subject have to raise their finger when a red
 shape occurrs.
 Result: As assessed by fMRI, oxygene levels were higer in the
 prefrontal cortex in the first condition. The effect was greatest
 in the right hemisphere.
 Conclusion: The right prefrontal cortex may play a role in memory
 retrieval. (see HERA)

Task tracer experiment (Goldman-Rakic 1994):
 Task: An animal is injected a slow radioactive tracer. The animal
 then performs a task and as a reward is killed.
 Result: The tracer shows the temporal sequence of brain activities.
 Conclusion: The prefrontal cortex does not store representations,
 it just temporarily loads them in order to perform a task.

Working memory PET experiment (Smith & Jonides 1994):
 Task: In the first condition, subjects have to judge whether a
 circle on the monitor encircles an area previously occupied by a
 dot. In the second condition, subjects have to tell whether a shape
 was shown before.
 Result: The first condition activated the right hemisphere, the
 second one the left hemisphere prefrontal cortex.
 Conclusion: The right hemisphere prefrontal cortex deals with
 spatial memory, the left hemisphere prefrontal cortex deals with
 object memory.

Recency discrimination experiment (Milner 1995):
 Task: Subjects have to study cards and are afterwards asked which
 of two cards was presented more recently.
 Result: Frontal lobe lesion patients have difficulties in
 accomplishing this task.
 Conclusion: The Frontal lobes seem to handle recency effects.
 See also: Inhibitory experiment.

Source memory experiment (Janowsky 1989):
 Task: Subjects are told common sense sentences. After a retention
 interval of some weeks, they have to answer common sense questions
 and tell where they acquired this knowledge.
 Result: Our poor Fronties performed well on the questions, but had
 difficulties in naming the source of the knowledge.
 Conclusion: The frontal lobe might be responsible for source memory.
 This task can also be done with healthy subjects with different
 frontal lobe strengths.

Highlighting theory (Shimamura 1995):
 The frontal lobes not only passively maintain representations, but
 they can also shed attention on details. The subject's goals trigger
 attention on certain aspects of the representations.

Inhibitory experiment (Knight 1995):
 Task: Subjects hear a sound on one ear and are instructed to either
 ignore it or pay attention to it. The brain potentials are recorded.
 Result: The brain potentials in both conditions differ
 significantly in normal patients, but not in prefrontal lobe
 lesioned patients.
 Conclusion: The prefrontal cortex filters information and inhibits
 unattended stimuli. Since the frontal lobe is responsible for the
 decay of information, frontal lobe patients have problems in the
 recency discrimination experiment: There is no fading away of
 memory which could be used to estimate the time.

Social interaction experiment (Lhermitte 1986):
 Task: A frontal lobe patient comes in, a needle is lying on the
 table, and the experimenter drops his trousers.
 Result: The patient makes immediate use of the needle :-)
 Conclusion: Frontal lobe patients have intact knowledge about the
 use of objects but they lack knowledge about social standards.
 This theory is also supported by the fact that animals which
 unfortunately "suffered lesions to their prefrontal cortex"
 (says the book) are outlawed by their group.

Emotion theory (Damasio 1994):
 Emotions play an important role in reasoning and thinking.
 Evidence: The prefrontal lobe is strongly interconnected with limbic
 structures tied to the emotions.

Somatic marker theory (Damasio 1994):
 There is an emotion-based decision mechanism which helps sorting
 possible options in a decision process. The somatic marker makes
 use of the emotional context memories are saved with. It thus
 constraints the playing field of possible reactions.

Elliot: A patient with frontal lobe lesions who did not have
 emotions. Consequently, his memories were stripped of emotional tags
 when they entered the prefrontal lobes and his somatic markers could
 not work, leaving him in a whirlpool of possible actions.

Galvanic skin response, GSR:  The electrical conductivity of the skin.
 Indicates emotional alertness.

GSR experiment (Damasio 1994):
 Task: Subjects view either neutral or shocking pictures. Their
 GSR is measured.
 Result: Normal subjects showed a GSR hike on the shocking pictures,
 but frontal lobe patients did not.
 Conclusion: Frontal lobe patients lack affective responses.

Risky card experiment (Damasio 1994):
 Task: Subjects uncover cards from two piles. Each card either means
 winning money or loosing money. One of the piles is risky (high
 absolute values), the other one is not.
 Result: Normal subjects had a high GSR when they opted to take a
 card from the risky pile while frontal lobe patients had not. The
 patients also preferred taking cards from the riskier stack.
 Conclusion: Emotions play a role in risk taking. Without the
 prefrontal cortex, behavior becomes coupled to stimuli of the
 present.

Goal experiment (Shallice 1991):
 Task: Three frontal lobe patients are asked to perform everyday
 business (e.g. go shopping).
 Result: None of them succeeded because they were unable to
 establish and achieve goals and subgoals.
 Conclusion: The frontal lobes are essential for planning and
 accomplishing goals.

Scheme control units theory (Norman 1986):
 The selection of an action is a competative process of "scheme
 control units". These units represent a behavior connected to a
 stimulus. The perceptual system provides the input for the
 scheme control units. Contetion sheduling takes care that
 executing two exclusive units is avoided. Furthermore, the
 supervisory attentional system (SAS) can explicitly select a scheme
 control unit. This is always necessary when we lack a routine
 procedure or an incorrect response is likely to occur.

Anterior cingulate theory (Posner 1994):
 The anterior cingulate is the central executive attention system
 (the SAS). It interacts with numerous other brain regions in order
 to ensure that processing in those regions is efficient with regard
 to the current task.
 Evidence: All situations which required the SAS activated the
 anterior cingulate, that is: Novel situations, error correction,
 overcoming habitual responses, difficult situations, decision
 making.

Two-choice letter discrimination task (Gehring 1993):
 Task: Subjects have to press one of two buttons according to one
 of two letters presented on the screen. Their brain activity is
 measured.
 Result: Whenever the subject makes a mistake and is aware of it,
 there is a potential in the anterior cingulate shortly after the
 action.
 Conclusion: The SAS (as located in the anterior cingulate)
 becomes aware of the mistake and tries to avoid it, but is too
 late.

Random finger experiment (Frith 1991):
 Tasks: Subjects receive a stimulus to either the first or the
 second finger of one hand. In one condition, they have to move the
 stimulated finger, in the other ("free-will-") condition, they were
 instructed to lift any finger.
 Result: The anterior cingulate was much more active in the second
 condition.
 Conclusion: The anterior cingulate coordinates the difficult response
 in the second condition, which requires much more working memory.

Frontal lobe, basal ganglia and cerebellum: Highly interacting
 brain systems, as evidenced by PET studies. Patients with lesions
 in the basal ganglia or cerebellum show impairments similar to those
 of frontal lobe patients. Cerebellum patients perform for instance badly
 in frontal lobe experiments. The cerebellum itself seems to be involved
 in verbal rehearsal and working memory, as can be assessed by the
 Wisconsin card experiment.