Lec 1/ TB Ch 1, 2 Flashcards

Question

* Neuropsychology cont 5 * Tumours * Definition * Glioma: definition * Growth rate * Kinno et al 2009 * Japanese-speaking patients w/ gliomas affecting Broca’s area * Method * Results * Using behavioral data and lesion data * 3 steps Lesion Overlap and Subtraction Analysis * Tranel and Kemmerer 2004 - understand neural basis of impaired knowledge on meanings of location prepositions * Locative prepositions * Part 1 * Methods * 4 tasks * Results * Part 2 * Method * Results * Voxel-Based Lesion–Symptom Mapping (VLSM) * 3 steps * Wu et al 2007 - VLSM & locative prepositions * 3 caveats About Neuropsychological Research on Structure–Function Relationships * 2 ways where areas are dysfunctional and may not show up on MRI * Apraxia of Speech (AOS) * Nina Dronkers 1996 * Method * Results * 3 limitations * Argye Hillis 2004 * Methods * Results * What went “wrong” w/ Dronker’s study?

Answer 1

* Tumours * Lesions can be due to tumors * Tumours: masses of tissue that grow abnormally and serve no physiological purpose * Several type (classified based on how they dev, and likelihood to recur after surgical removal) * 1. Glioma * Come from white matter and expand outward -\> destroy or displace neurons * Variable rate of growth (can be v slow or fast) * Glioma patient studies are useful * Kinno et al 2009 * A group of Japanese-speaking patients w/ gliomas affecting Broca’s area had sig worse comprehension for passive sentences than active ones * Ex. Passive – Person A is being pushed by Person B * Ex. Active – Person A is pushing Person B * Original stimuli: Japanese sentences paired w/ scenes with 2 ppl * Another study * Showed these patients’ lesion sites overlapped w/ an area that was engaged sig more by passive than active sentences based on another fMRI study * They compared them to healthy controls who did the same task * Since the neuropsychological data and fMRI data aligns, this shows reliable inferences can be drawn from Glioma studies Relationships Between Behavioral Data and Lesion Data * -\> discuss how groups studies are conducted * Lesion Overlap and Subtraction Analysis * 1. Two groups of patients * Gp A w/ deficit of interest * Gp B do not have it * 2. Each groups has separate procedures * Reconstruct the patients’ lesions on a standard brain template * Calculate how much overlap for every voxel * 3. Lesion overlap map for patients w/ deficit minus lesion overlap map for patients w/o deficit -\> remaining area = areas of damage that is linked w/ deficit * Tranel and Kemmerer 2004 * Tried to identify the neural basis of impaired knowledge on meanings of location prepositions * Locative prepositions: * Ex. in, on, around, through, above, below * Methods * 4 tasks given to 78 brain-damaged patients (from stroke) * 1. Naming (80 items) * Ppl are shown objects and are asked to orally name the location of one object relative to another * 2. Matching (50 item) * Ppl are shown 3 spatial arrays of objects w/ a preposition * Told to choose which array best represent the preposition * 3. Odd one out (45 items) * Ppl are asked to choose which one involves a type of relationship that is different from the other two * 4. Verification (44 items) * Ppl are shown a spatial array of abstract shapes together and a preposition * They need to decide whether the preposition correctly describes the array * * Results * Very variable: some failed none, some failed one task, some failed all 4 tasks * Those who failed all 4 tasks may have impaired knowledge on the location preposition meanings * Those who only failed only 1 task had unimpaired knowledge on meanings but may have individual disturbances on certain processes * Part 2 * Rs created lesion overlap maps for those w/o impairments; and one for those w/ impairments * Then they subtract the maps from each other * Result shows defective knowledge on the meanings of location prepositions is associated w/ damage in some areas in LH * The cortex and underlying white matter in the inferior frontal, inferior parietal, and posterior superior temporal regions (red, orange, yellows) * * The analysis includes patents w/ the deficit of interest and those w/o * So, the results provide strong evidence that the lesions mentioned are more likely than not to cause the deficit * Voxel-Based Lesion–Symptom Mapping (VLSM) * 1. Administer a task to patients w/ various lesions * 2. Patients’ lesion sites are transferred to a standard brain template * For each voxel, 2 gps of patients are formed: * Those w/ brain lesions vs those w/o * 3. For each voxel, rs do a t-test to compare the behaviour performances of patients in the lesioned and non-lesioned groups * The resulting t-values measures the degree on whether decreasing task performance is related to the presence vs absence of certain brain areas being damaged * IOW, VLSM allows rs to identify the neural reasons of deficits, quantify brain-behaviour relationships w/o classifying patients based on their impairment * Wu et al 2007 * Used VLSM to study impaired knowledge of meaning of location prepositions * 14 patients w/ LH lesions * They do several tasks the sentence w/ location prepositions w/ 1 of 4 pics * Results: accuracy range b/w 43% to 100%’ mean = 86% * The lesioned areas are consistent w/ those from Tranel and Kemmerer (2004 – abv) * A Few Caveats About Neuropsychological Research on Structure–Function Relationships * 3 caveats * 1. To infer a causal relationship b/w an impaired ability and damage to a specific region, need to show that * Patients w/ deficit have lesions at that site * Patients w/ lesions at that site tend to have the deficit * 2. Deficit-lesion correlations for those w/ stroke/TBI in the chronic period can be diff from those in the acute period * Chronic period: 6+ mo after lesion onset * Acute period: less than 6 mo after lesion onset * This is b/c acute patients w/ small lesions recover fast; the intact areas “take over” the affected fx * 3. Some impairments may be due to areas that are not structurally damaged * These areas are dysfunctional and may not show up on MRI * 2 ways * A. Hypoperfusion: receives enough blood supply to survive but not enough to operate normally * B. Diaschisis: they depend on axon input from the site of structural damage; but the input is no longer available * Apraxia of Speech (AOS): disorder of articulatory programming in Broca’s aphasia * AOS affect high-level motor aspects of orchestrating speech * This leads to distorted consonants, vowels, and prosody * Patients know what they want to say and how it should sound, but cannot coordinate the articulators (i.e. lips, tongue, jaw, and palate) to produce the output * Nina Dronkers 1996 * Showed 25 chronic stroke patients w/ AOS had 100% lesion overlap in one LH region – superior anterior part of the insula * 19 chronic stroke patients w/ AOS had that damage * 3 limitations (related to 3 caveats above) * 1. Results show that AOS is associated w/ a damaged area of insula; but it doesn’t tell if damage to that area give rise to AOS * It is possible that AOS may also involve damage in other areas * 2. All the patients were chronic; acute patients may show diff deficit-lesion correlation * 3. The study only focused on structural damage * Other areas that are intact but are dysfunctional by contribute to AOS * Argye Hillis 2004 * Addressed the 3 limitations in Dronker’s study * Methods * 1. Selected 80 stroke patients based on site of structural damage * 40: lesion on any part of the left insula * 40: spared that area * 2. Patients were eval for AOS when they were acute (24 hr of stroke onset) * 3. Patients were scanned of tissue damage, and also areas of hypoperfusion (sig reduced blood flow) * Results * Abnormalities in region of interest (superior anterior part of left insula) is not sig related to AOS * 12 out of 29 patients (40%) who had damage or hypoperfusion in that area showed AOS * 19 out of 51 patients (40%) who did NOT hv damage or hypoperfusion in that area showed AOS * AOS is actually associated w/ abnormalities in Broca’s area * 26/30 (90%) who had damage or hypoperfusion in that area showed AOS * 5/50 (10%) who did NOT hv damage or hypoperfusion in that area showed AOS * What went “wrong” w/ Dronker’s study? * The insula is the most vulnerable to stroke due to anatomy * Maybe 80% of Dronker’s chronic patients had damage to Broca’s area, while the remaining patients had hypoperfusion in that area

Answer 2

**_Functional Neuroimaging_** * Aka hemodynamic methods (b/c vampire theory) * PET & fMRI (more common) * Map the function in human brain * Vampire theory: the more active regions suck more blood * Key process * 1. When the cog task uses a particular brain area, the active neurons in that area rapidly consume the oxygen available in the local capillaries * 2. A few seconds later, there is an overabundance of oxygenated blood * Fulton 1930s * Dr. Fulton worked w/ Walter who had many congenitally abnormal blood vessels (arteriovenous malformation) in occipital cortex * It causes headaches and visual disturbances * When blood pass through those vessels intensely, it produces a pulsing sound that the patient hears as a humming sound * Dr. Fulton can also hear it when putting a stethoscope at the back of patient’s head * Dr. Fulton found that the rushing sound is correlated w/ Walter’s heartbeats and visual experiences * Ex. If Walter lied in the dark for a few minutes and suddenly use his eyes OR reading smth attentively, the sounds gets louder * This shows that changes in blood flow reflect changes in neural activity in certain regions, and this affect cog changes Two Techniques * Positron Emission Tomography (PET) * Measures regional cerebral blood flow, which infers regional neural activity * The machines tracks the distribution of radioactive isotopes throughout the brain * 1. Rs create an isotope in a medical cyclotron, and place it in water * 2. This radioactive water is sent to the imaging facility, and rs inject it into the subject’s bloodstream. The bloodstream carries the radioactive water into the subject’s brain. * 15 O is the most common isotope used * It has 8 p+ and 7 n0 * When it decays, it releases a proton that is +vely charged (i.e. positron) * The positron travels and is attracted to surrounding/ambient electron; this causes the electron and positron to be destroyed (annihilated) * As a result, 2 photons are released in opposite directions ![]() * These photons exit the head and are picked up by the rings of detectors in the PET scanner * The computer reconstructs the annihilation location in the subject’s brain * If there is a large # of annihilation events in an area, this suggests there is more blood flow in that area. This in turn indicates there is more neural activity * Spatial resolution: 10mm (kinda bad) * Temporal resolution: 30s (bad) * Reason: need to average the data over the length of time to get a good signal-to-noise ratio??

Answer 3

* Functional Magnetic Resonance Imaging (fMRI) * Sensitive to degree of oxygenation of blood in diff parts of brain * Blood oxygenation level dependent signal (BOLD) * Endogenous contrast agent: deoxygenated blood reduces the MRI signal; oxygenated blood increases this * When hemoglobin is not carrying any oxygen, it behaves like a small magnet and disrupts the local magnetic field * This reduces the signal received by the fMRI machine * When hemoglobin is carrying oxygen, it is less magnetic * The fMRI can still detect them as a small increase in resonance signal * 12 second hemodynamic response function * 1. When neurons in a specific area increase in activity, they consume the oxygen that is immediately available * 2. As a result, there is more deoxygenated blood in that area, and the fMRI will pick up a small decrease in BOLD signal for 2s * 3. In the next 5s, more oxygenated blood is detected in that area, and the fMRI detects this as a gradual increase in the BOLD signal * 4. For the next 5s, the signal will return to baseline and briefly undershoots * ![]() * The BOLD signal maybe more correlated to the input processing of neurons rather than output * Spatial resolution: 1-3 mm * Temporal resolution 50-100 ms * Pros of fMRI over PET * 1. Cheaper: doesn’t need a medical cyclotron * 2. Safer: doesn’t use radioactive isotopes * 3. Better spatial and temporal resolution * Cons * 1. It is very noisy * 2. BOLD signals are distorted near air-filled cavities * It is tricky it get data from some brain regions, esp the anterior temporal lobes located near the sinuses * (rs found a way around this)

Answer 4

Standardized Three-Dimensional Coordinates for Defining Stereotaxic Brain Space * In fMRI studies, rs transform the anatomic confiscation of each subject to a standard brain template, then they combine data across subjects * Stat sig activations are identified and reported in the common stereotaxic space of the standard brain template * The brain template uses the x,y,z coordinate system * The origin is the anterior commissure * * X-axis: right-left dimension (right = +ve; left = -ve) * Y-axis: anterior-posterior (anterior = +ve; posterior = -ve) * Z-axis: superior-inferior (superior = +ve; inferior = -ve) * 2 diff atlases/3D grids (unit: voxels) * 1. Talairach and Tournous * Based on anatomical data from 1 post-mortem brain * 2. Montreal neurological institute (MNI) 1994 * Based on MRI scans from 305 healthy subjects * They are similar but not identical * Old studies voxel size: 3-5 mm * New studies: 1 mm * Potential cons of analysing multiple subjects for fMRI * Fedorenko and Kanwisher 2009 * This may obscure indiv diff in neural organization of language by removing anatomical variability b/w ppl’s brains * Potential solution: Plot each subject’s unique functional data on his/her anatomical data * 1. Locate a region of interest (ROI) in each subject * Ex. Fusiform Face Area – sensitive to seeing faces * 2. Test for other response properties for that ROI * Ex. See if FFA is more sensitive to upright vs inverted faces * 3. Conduct analyses across subjects using data from corresponding functional regions rather than drawing from the same locations in Talairach/NMI space * May provide more insight

Answer 5

Blocked Versus Event-Related Designs * How should we order individual trials in experimental condition? * Ex. You want to identify and compare brain regions that respond to words for animals (ex. rabbit) and words for tools (ex. pencil, knife) * -\> How do we sequence the trials for animal words and tool words? * This depends on the imaging technique * PET * Need to use block design b/c PET has poor temporal resolution * Block design * Trials for the same experimental condition are grouped together in blocks * Here, the subjects must do the same type of task for the same type of stimuli throughout the block * Ex. You set up 3 blocks in your PET study (each block = 1 min) * Block 1: Condition A – ppl read words for animals * Block 2: Baseline condition – ppl lie in the scanner; do nothing * Block 3: Condition B – ppl read words for tools * These blocks are separated by 10 min periods – this is to allow time for the radioactive isotope subjects receive at the beginning of each block to decay * Hypothetical results: * Strongest signal: Block 1/Condition A- read words for animals * Weakest signal: Block 2/Baseline – rest * In b/w: Block 3/Condition B – read words for tools * * fMRI * Since it has better temporal resolution, you have more design options (2 options) * Option 1: You can use block design, but there will be * More blocks per condition * Shorter blocks (30s) * Same hypothetical results: * Strongest signal: Block 1/Condition A- read words for animals * Weakest signal: Block 2/Baseline – rest * In b/w: Block 3/Condition B – read words for tools * * Option 2: Event-related design * The trials belonging to diff experimental conditions are randomly scattered * Each trial has its own hemodynamic response * These indiv signals are later extracted and averaged based on the condition * Same hypothetical results * Strongest signal: Condition A- read words for animals * Weaker singal: Condition B – read words for tools * * Why are event-related design popular? * 1. Allow rs to randomize stimuli * Reduces the chance the results are influenced by habituation, anticipation, or strategic processing (which r more likely in block design) * 2. There’s more experimental flexibility than block design * Trials can be sorted by diff criteria * Ex. conditions, uncontrollable factors (trials subjects responded incorrect) * 3. Can see more details on amplitudes and time trends of hemodynamic response in diff brain regions

Answer 6

Some Basic Experimental Paradigms * Subtraction, correlation, and pattern analysis are mainly used w/ event-related design, but occasionally w/ block design * Subtraction paradigm * First paradigm used in fx neuroimaging * Can be used with blocked and event-related designs * Used by fMRI and PET rs * Helps rs isolate neural correlates of specific cog capacities * * Study design: 2 conditions * Experimental: ability of interest * Baseline/Control: does not include ability of interest, equivalent in other aspects * 1. Collect imaging data * 2. Map of brain activity w/ experimental condition “-“ map of brain activity w/ control condition * The regions that still show up is sig engaged in the experimental condition; this is treated as uniquely contributing to the ability of interest * To enhance reliability, we include * multiple experimental conditions that require the ability of interest (Ex. conditions A and B) * multiple control conditions that do not (Ex. C and D) * This allows rs to conduct conjunction analyses that increase the likelihood of pinpointing brain regions responsible for the interested ability * * Ex. If each experimental condition is well-matched w/ a particular control condition * Ex. A w/ C, B w/ D * 1. Perform separate subtractions b/w corresponding conditions * 2. Conjoin outcomes of the independent analyses [(A – C) + (B – D)] * The resulting brain map is quite restricted, and helps narrow the neural underpinnings of specific cog capacities * Narain et al 2003 * Aims to disclose brain regions linked w/ auditory processing of intelligible vs unintelligible sentences * 4 conditions: 2 using intelligible stimuli, 2 using unintelligible stimuli * Intelligible conditions * A. Speech (Sp): normal sentences * B. Vocoded speech (VCo): sentences that are artificially altered to have a rough sound quality but still can be understood * Unintelligible conditions * C. Rotated speech (RSp): Sentences that are spectrally rotated (i.e. inverted) so they sound like “alien” language; * but still have phonetic features (ex. voiceless fricatives – “f”, “th”) * D. Rotated vocoded speech (RVCo): Sentences that have been vocoded and spectrally rotated (i.e. inverted) so they sound like intermittent fluctuating static * 11 ppl listened to blocks of stimuli from each condition during fMRI scanning * All state they can only understand stimuli 2 out of 4 conditions * Rs identified the brain regions that mediate extracting meaning from intelligible utterances using the “subtraction” method * [(A – C) + (B – D)] * [(Sp – RSp) + (VCo – RVCo)] * 1. Do 2 separate subtractions * Each involves contrasting the activation evoked by intelligible stimuli against the matched set of unintelligible stimuli * Intelligible and unintelligible conditions differ only by spectral rotation * 2. Conjoin (add) activation maps from the 2 separate subtractions; this identifies the commonality * This captures brain areas that respond to both intelligible stimuli * This rule out areas that respond to only one intelligible stimuli and areas that respond to both unintelligible stimuli * Final activation map * Only 3 clusters of voxels are stat sig * They are in the left lateral temporal lobe * Some portion overlap with Wernicke’s area, * Damage to that area disrupts phonological and lexical aspects of spoken language comprehension * It’s possible lateral temporal region contributes to high-level auditory sentence processing, not low-level acoustic and phonetic info * Note: analysis was restrictive * Correlation paradigm (aka parametric paradigm) * Compatible w/ blocked and event-related design * The ability of interest here is a continuous variable that can vary; it is not all-or-nothing * Subjects perform a series of tasks that uses the ability of interest in diff extent * Rs look for brain regions where the activity level changes correspondingly * Mobbs et al 2010 * Looked at the neural activity associated w/ monitoring the threat value of a tarantula * 1. Ppl were placed in fMRI, one foot is in a box w/ 5 compartments * 2. Ppl think they were seeing a live camera feed where a rs move a large tarantula from one box to another, closer/farther from their foot * Reality: they are liking at a pre-recorded clip * Results from parametric regression analysis * As the spider got closer to the foot, brain activity increased in anxiety-related areas * Dorsal anterior cingulate cortex, midbrain nuclei * As thee spider got farther from the foot, the brain activity progressively increased in areas that are linked w/ passive coping w/ distant dangers * Anterior orbitomedial PFC * This reveals the neural circuitry for arachnophobia * Davis and Johnsrude 2003 * Used correlation paradigm to further explain the neural underpinnings of intelligible and unintelligible * Intelligibility is on a continuum * 1. Used fMRI, subjects listened to diff stimuli * Undistorted condition * A: Speech: normal sentences * Partially distorted conditions * B: Vocoded speech: sentences altered to sound rough but intelligible (similar to Narain et al 2003) * C: Segmented speech: sentences altered by dividing the speech stream into separate chunks and replace some parts w/ signal-correlated noise * These parts retain many original acoustic features but lack recognizable sounds * D: Speech in noise: sentences are placed in the background of continuous speech spectrum noise? * Completely distorted condition * E: Signal-correlated noise: sentences altered in the same way in the “segmented speech” condition * Here, all of the chunks are replaced w/ signal correlated noise -\> so the stimuli is completely incomprehensible * NOTE: in the 3 partially distorted conditions, there are 3 diff degrees of distortion leading to 3 diff degrees of intelligibility * Based on pilot study * High intelligibility: 90% of words reported correct * Medium intelligibility: 70% of words reported correct * Low intelligibility: 20% of words reported correct * 2. The subjects pressed buttons to rate the intelligibility of each stimulus on 4-point scale * Results of analyses * Subtraction paradigm: compared distorted condition w/ silence * Sig bilateral activation in Heschl’s gyrus and surrounding auditory cortices * Corelation analyses: see which brain regions respond w/ gradual increase in BOLD signals when the intelligibility of the stimuli increases * The following regions show Neural sensitivity to cont variation in intelligibility of stimuli: * 1. superior and middle temporal gyri in LH; extends anteriorly to temporal area and posteriorly angular gyrus * 2. Similar to #1 but distributed more anteriorly in RH * 3. Small part of Broca’s area * Follow-up analyses * There is a single voxel in the left anterior temporal lobe that was sensitive to intelligibility * Activated strongly to undistorted condition * Activated weakly in completely distorted condition * Activated w/ varying intensities across 3 partially distorted conditions depending on the amount of distortion * (i.e. 90%, 70%, or 20% intelligible) * IOW: the voxel is not sensitive to the diff ways they are distorted (vocoded, segmented, or embedded in background noise) * Take away from this study: brain mechanisms for comprehension of intelligible utterances involve the LH areas mentioned in Narain et al 2003’s study * It also involves the RH * Here, the correlation paradigm allows rs to discover more brain areas involved * Gradual changes in the stimuli taps into diff brain regions * Multivariate Pattern Analysis * A new paradigm that can decode more fine-grained patterns of brain activity to more precision than the other paradigms * In the subtraction and correlation paradigms, each voxel is independent, and the analyses shows separate measures of signal strengths * When we use these paradigms to look at the whole brain, it looks at the mean response of each isolated voxel in each condition * When we use it to look at the region of interest (ROI), it spatially averages the signals from all the voxels in the region for each condition * Limitations: this ignores the relationship b/w voxels, and this can’t detect any info that is latent (dormant) * MVPA can identify these dormant patterns, and relate them to cog abilities/tasks * Mur et al 2009/ Raizada et al 2010 * MVPA can help explore whether the perception of similar syllables (ex. ra and la) is associated w/ distinct levels or patterns of brain activity in the superior temporal ROI spanning 9 voxels * If the stimuli triggered orthogonal (right angle) 4-voxel activation patterns in ROI, the conventional analysis would miss this * Conventional analysis would generate output that shows the same activity for both speech sounds and suggest that ROI cannot discriminate b/w the two * MVPA can expose this separate pattern, and suggest the ROI contains at least partially non-overlapping neural voxels that represent the 2 syllables * Point: MVPA is very useful * Abrams et al 2013 * Used MVPA to examine the neural substrates that process intelligible vs unintelligible utterances w/ better precision * 1. Present 20 ppl w/ 2 types of stimuli, sequenced in alternating blocks * A: Speech (Sp): Intelligible excerpts from famous speeches in 20 C * B: Rotated speech (RSp): same experts but spectrally rotated (inverted) as in Narain’s 2003 study * 2. Ppl play close attention to each excerpt and push a button when it ended * Rs conducted 2 analyses: Subtraction and MVPA * I. Subtraction paradigm/General Linear model (A – B) -\> (Sp – RSp) * II. MVPA, searchlight version * For every voxel in the brain, a 3 x 3 x 3 neighbourhood is made, centred in the given voxel * W/in the 27-voxel space, rs determined is they were diff * Results * Red: conducted subtraction/General linear model (GLM) analysis * Areas include 3 left lateral temporal regions (as seen in Narain et al 2003) * Most of the bilateral temporal regions (as seen in Davis and Johnsrude 2003) * Green: MVPA/ searchlight analysis; Also detected by GLM analysis * There are 2 patches where the multi-voxel patterns of activity were sig diff for the two conditions * Blue: MVPA/ searchlight analysis; NOT detected by GLM analysis * Picked out RH areas the can discriminate b/w the two conditions * Located in temporal, parietal, and FL * They are associated w/ auditory sensory processing * * Conclusion: results suggest MVPA can resolve inconsistencies in the literature by being more sensitive in data analysis

Answer 7

**_Electrophysiology_** * Both PET and fMRI does not directly measure brain activity; they only track the metabolic consequences * 2 Electrophysiological approaches * Directly stimulate specific parts in the brain and observe the effects on cognition and b * Record electrical signals of neurons * 1. Intracranially: put electrodes in the brain to record the firing of single cells or a cluster of cells * 2. Extracranially: put electrodes n the scalp to record cells firing Stimulation * Penfield 1956 * Direct electrical stimulation to map the functional organization of the cortex * Did this on patients’ w/ epilepsy before removing the brain areas that were thought to cause seizures * Used local anaesthesia only as there are no pain receptors in the brain * Found that stimulating some areas induce “+ve responses” * Ex. stimulate PMC cause involuntary movements in specific body parts * Ex. Stimulate Primary somatosensory cortex evoked feelings in specific body parts; they correspond to the somatosensory homunculi * Ex. Stimulate occipital regions cause visual hallucination (color, starts, light flickers) * Ex. Stimulate superior temporal regions cause auditory hallucinations (ringing, clicking, buzzing) * Ex. If he stimulated higher-order temporal regions, this replays the patient's past experiences that were richer than ordinary recollections * flashbacks/dreams * This is rare * Stimulation to language-related areas can induce “-ve responses”, esp for speech * Speech arrest: slowing of speech production * Anomia: can’t retrieve a word * Paraphasia: distort the semantic or phonological content of a word * Ex. say “tephelone” instead of “telephone.” * This showed that stimulating specific sites and cause linguistic errors * Ojemann 1989 * directly stimulate language-related areas to study language * mapped out the “eloquent” cortex in the left perisylvian territory * 120 patients w/ epilepsy * 1. Showed line drawings of familiar objects on a screen in 4-sec intervals * 2. Patient had to name each one using the carrier phrase “this is a \_\_\_” * 3. At the beginning of some slides, the rs stimulate one of the several predetermined points on the exposed cortex * Each part was stimulated 3 times; w/ all sites were stimulated once b4 any of them were stimulated a 2^nd time * 4. Rs provide immediate feedback on patient’s response * If stimulating a given site induced error for 2 out 3 trials, that site is essential for naming * 2 main findings * 1. For most cases, stimulation disrupted naming as a few discrete sites * In 70% patients, 2+ of these areas were identified: 1 in FC, other in temporal/parietal cortex * 2. Rs did stimulation mapping on 90 patients in anterior and posterior perisylvian zones * 20% had frontal naming sites * 15% had temporal and parietal naming sites * Main point: precise naming sites in the brain vary across patients * There was only 1 site where errors were evoked more than 50% of the time (located in posterior part of Broca’s area) * Some say electrical stimulation is the “gold standard for brain mapping” * Cons: * 1. we don’t fully understand the physiological cons of passing current through the cortex * 2. The intensity of stimulation that is needed to identify “eloquent” areas differ across patients and regions * 3. Stimulation to a given site can propagate to other remote brain regions * Need to interpret these findings w/ caution! Recording * Intracranial * Some rs record neural activity in particular regions * It helps doctors to precisely localize the source of epileptic seizures * Procedure * 1. Implant electrodes in the tissue that is suspected to surround the epileptogenic site * 2. close the patient’s head and record neural activity for several days cont * If the seizure occur, we can identify the point of origin * Some rs determine whether and how the firing rates of neurons in the implanted tissue are modulated (regulated) * They study the neural activity in single cells of cell assemblies * Creutzfeldt et al 1989 * Found that individual neurons in the right superior temporal gyrus are sensitive to certain features of auditorily perceived words (ex. phoneme, syllables, morphemes) * Ex. when the patient listened to a list of multisyllabic words, the firing rate of one cell increased sig in response * The critical sounds are velar consonants (i.e. /k/, /g/) combined w/ /r/ or /s/ * Ex. Christmas, crocodile * When the entire list of words was presented again after 2 min, the cell’s response to each word was similar to the response in the 1^st time * Local field potentials: the sum of the extracellular voltage fluctuations that indicate the sum of events in the dendrites for a local population of neurons * This is measured when the patient is doing linguistic tasks * Methods * Option A: Use electrodes that penetrate many layers of cortex and white matter/ or target subcortical nuclei * Option B: use high-density multi-electrode grid, place it directly over the cortical surface to collect data (aka electrocorticography) * Flinker et al 2011 * Studied cortical dynamics of speech perception in 3 epileptic patients * 1. Placed an electrode grid on patient that covers the posterior lateral surface of the left superior temporal gyrus * 2. Ppl listened to 2 types of stimuli * I. Syllables (for 150 ms): /ba/, /da/, /ga/ * II. 3 kinds of words spoken by the same talker * 23 pseudowords (3 phonemes) * 23 real words (3 phonemes) * 4 proper names (5 phonemes) * Results: * Electrode A (top row): high frequency neural activity response only to words * Electrode B (bottom row; 4mm away to electrode A): high frequency neural activity response to both words and phonemes * Most activity starts after 200 ms * Point: electrocorticography is useful * Extracranial * Electrical potential produced by large populations of neurons are strong enough to be conducted passively though tissues of the brain, skull, and scalp * This is a method to record patterns of activity in a non-invasive manner * You place electrodes on the surface of the scalp and compare the voltage fluctuations here w/ a reference location (ex. mastoid bones behind the ears) * Electroencephalogram: recording on the changing electrical activity; data is from an electrode on the scalp * Has rhythmic undulations (ups and downs) that vary in amplitude, duration; it depends on whether the subject is excited, relaxed drowsy or in a stage of sleep * EEG is a collection of many diff neural activity; so it doesn’t provide info on mental operations * Rs examine the neural correlates of specific cog capacities * Rs examine how the EEG recorded at diff electrodes change when ppl are doing the tasks * Event-related potentials (ERPs): EEG patterns that are measured when the stimuli is presented * 4 ERP parameters * 1. Latency: time-point, milliseconds after the stimulus; happens when a particular deflection of waveform begins/reach its peak * 2. Amplitude: strength of an effect (mV) * 3. Polarity: whether a deflection is +ve-going or -ve-going * 4. Topography: the scalp distribution of an effect; involved the electrode positions at where the effect was observed * Some labs plot -ve up and +ve down OR vv * When a wave deviates +vely or -vely, it usually is not sig * This is b/c polarity is affected by many irrelevant factors * Ex. location of reference electrodes * Ex. orientation of the source of scalp recorded signals * Ex. if a batch of cortical neurons have their dendrites point to the surface and simultaneously get excitatory input, the resulting electrophysiological activity create current “dipoles” * Each dipole has a -ve end near the dendrites and a +ve end near the cell body * The summation of these dipoles is detected at the scalp as a -ve voltage * This does not mean the operations are done by the neurons, it’s just activity on detected on the scalp * Since most ERP studies compare waveforms from the experimental to control condition, the diff in polarity maybe related to cog processes * The key thing is there is a polarity diff; the +ve/-ve doesn’t matter * Topography * Studies vary on the # of electrodes used and their size * As the size increases, the difference of ERP effects along the scalp increases * The larger range of channels, the easier it is to apply data analysis to somewhat infer the neural underpinnings of ERP effects * Inverse problem: ERP has poor spatial resolution but v good temporal resolution * Another issue: the electrical activity recorded at the scalp can come from any neural generator * N400 * N = negative * 400 = a peak that appears 400 ms post-stimulus * centroparietal electrode can detect a N400 w/ the greatest amplitude * N400 can track the gradual build-up of semantic content during receptive sentence processing * The amplitude increases as we increase the difficulty of integrating the meaning of a word into the preceding context * Kutas and Hillyard 1980 * Showed the waveforms evoked by 3 diff types of sentences presented to subjects on a screen, 1 word at a time * Condition 1: all sentences are normal * Ex. It was his first day at work * Condition 2: all of the sentences ended w/ a word that was semantically anomalous in the given context * Ex He spread the warn bread w/ socks * Condition 3: all the sentences ended w/ a word that was orthographically incongruent in the given context * Ex. She put on her high heeled SHOES * Results: final word in anomalous condition trigger a large N400 * This indicates there was effortful conceptual processing, not surprise * The final word in the orthogonal condition is associated w/ a P560 *

Answer 8

**_Box 2.2: Magnetoencephalography_** * Non-invasive brain mapping technique related to EEG/ERP * Electrical currents associated w/ neural activity create tiny magnetic fields, these fields are recorded at the scalp and time-locked to when the stimuli is presented * This produces event-related fields (ERFs) * In MEG systems, we detect ERFs w/ 200 sensors called SQUIDS * SQUIDS: superconducting quantum interference device * Pros of MEG/ERF * Outstanding temporal resolution (in ms) * Provide good spatial resolution, can help neurosurgeons identify where the seizure happens in epileptic patients * This is possible b/c magnetic fields are not distorted as they pass through the brain, skull, and scalp * Magnetic field strength dips from their source in a systematic manner * Cons * It is sensitive to neural activity in the gyri but NOT the sulci * Too $$$ * This is b/c ERFs are tiny signals that are a billionth the size of earth’s magnetic field * The lab has to be heavily shielded from outside magnetic fields * SQUIDS need to be stored in liquid helium * More studies are using this despite the cost

Answer 9

**_Transcranial Magnetic Stimulation_** * It is very invasive and restricted to those w/ a history of nro dysfunction -\> TMS is not used often * TMS can quickly alter the organization of neural activity in certain cortical regions by placing magnetic device on the scalp * You can facilitate or supress the operation of the target area * If safety protocols are followed, TMS is harmless; w/ may cause seizures * It can help offset clinical disorders like stroke-induced aphasia, and study language/MD How It Works * Uses EM iniduction * Faraday * When an electrical current passed thru a wire, it generates a magnetic field that varies w/ time * If wire 2 is placed nearby, the magnetic field induces an electric current flow in wire 2 * For TMS, “wire 1” is the stimulating coil”; “wire 2: is the targeted region of the brain * Common coil type: figure-8 shape, current flows around 2 adj parts and sum up at the intersection * The flux/magnetic field lines are perpendicular to the plane of the coil * * When the central part of the coil is placed at a predetermined position on the scalp, the magnetic field that passes through the skull * this temp changes the electrophysiological properties of the neurons under the cortical region * * Temporal resolution: ms * Spatial resolution: mm * Shifting the coil b/w areas separated by .5 to 1 cm over the PMC can trigger muscle twitches that follow the motor homunculus layout * Rs use the intensity threshold for triggering a finger twitch as a benchmark for calibrating the strength of pulses delivered to other regions for high-lv cog processes * 2 strategies are used to determine the appropriate position of a coil (can use 1 or both) * 1. Functional localization: move the coil around a general territory until the ability of interest if enhanced/disrupted * 2. Anatomical localization: use neuronavigation system to guide the placement of the coil * based on data from structural or fMRI OR a set of standardized coordinates * Issues: effects of TMS can spread to remote brain regions; there may be other b cons * * There are many factors that induce facilitatory or inhibitory effects * One main parameter: frequency of pulses * Ex. 2 single pulses separated by less than 5 ms -\> intracortical inhibition * Ex. 2 single pulses separated by 10 – 30 ms -\> intracortical facilitation * Some pulses can be applied to subjects “online” (during the task performance) * Some pulses are applied “offline” (prior to task performance) * Ex. Common technique – apply repetitive pulses to target site cont for several min b4 task * Reduce excitability of that region after stimulation * This can affect subsequent performance * TMS cons are measured in RT or accuracy Applications to Language * TMS helps us understand more about language-related brain regions * Gough et al 2005 * Used TMS and found the complementary roles of 2 diff sectors in the left interior frontal gyrus (LIFG) in making semantic and phonological judgements * Rostral (pars orbitalis) * Caudal (pars opercularis) * Subjects did 3 task w/ visually presented pairs of letter strings * Synonym judgement – ex. decide if idea and notion hv the same meaning * Rhyme judgement – ex. decide if eye and fly end w/ the same sound * Visual judgement – ex. decide if txbufr and txbufr have the same characters * There are 15 pseudo-randomly organized blocks of 10 trials, w/ the trials in each block involving the same type of task * 40% of the trials * There are 3 TMS pulses separated by 100 ms and starting 100 ms post-stimulus onset * These were delivered to the rostral or caudal of LIFG * There are equal # of TMS trials for each task; no more than 2 TMS trials consecutively * Results * Double dissociation emerged b/w semantic and phonological tasks based on the function of the site of stimulation * Judgements on the meanings of word pairs are delayed by rostral LIFG stimulation, not caudal LIFG * Judgements on the sound of word pairs were sig delayed by caudal LIFG stimulation, not rostral stimulation * Judgements on the appearance of letter strings were not affected by either site * There are 2 target areas in the LIFG that make diff contributions to lexical processing * These areas are close together; TMS can modify brain activity of brain areas that are close together *

Answer 10

**_What is language_** * Examples of communication (i.e., there needs to be some interaction between multiple individuals, usually of the same species, to communicate info): * Ex. Zebra Finches: male communicating (mate calling) with a female * The sounds are sequences of predictable sounds (aka motives) * There is enough similarity b/w Zebra finches and humans when communicating * It lacks some aspects of communication (ex. we don’t know what is being said) * Monkey Alarm Calls: aka notifying each other of a predator * Monkeys have different alarm calls for different predators * Alarm call 1: leopard alarm call * Monkeys from different species recognize each other’s alarm calls * They will move closer b/c once the leopard knows it has been spotted, it will give up * Unlike the zebra finch, we know what is being said (leopard is here) * Communication here is more sophisticated * These are not perfect examples of language * Main point: there is a continuum of communication; the more complex form is used in humans

Answer 11

**_Neural basis of language_** * The brain’s language network is formed by a combination of areas (networks) of brain regions that have become differentially specialized for language, but that clearly also relate to other basic cognitive abilities * Different systems are used in different aspects of language; these brain areas are also used in other functions (non-language) * Vision (ex. reading) * Audition (ex. listening) * cross-modal associations * motor outputs (ex. writing, Brail) * planning/executive control (ex. produce a response) * ![]() * How do we understand words? * 1. Extract words from auditory segment * 2. Separate them into phonemes * 3. Associate these to words * 4. Put words together to determine its meaning * 5. Produce a response * If certain language functions are present in other species (like birds and monkeys), these functions may evolved to be more specialized in humans

Answer 12

**_A few (of many!) properties of language_** * “words” convey “meaning” (lexical semantics) * Ex. monkeys has a specific alarm for leopards (basic meaning) * Grammar, morphology, and syntax (particular words in particular order) * Ex. not present in monkey language (subject, verb, object structure; location of leopard) * Generative (i.e., we can create new sentences that everyone can understand) * Ex. not seen in monkey speech * Specialized for communication * Ex. human language is used for communication specifically * Ex. Hammer is not specialized for communication – can be used as an alarm, or banging on a pot for music * BUT, it stills builds upon other basic abilities like vision, audition, and motor systems, to name a few * Other important properties?

Answer 13

**_Cognitive Neuroscience of Language_** * Fundamentally, an interdisciplinary domain, comprised of at least 3 major domains: * Cognitive psychology * Neuroscience (Neuropsychology = precursor to NRO) * Neuropsychology: post-mortem study on what aspects of the patient is impairing language * Linguistics * Although many other areas make their appearance as well: * Evolution / Cross-species comparisons * Behavioral, neural levels * Some rodents can detect similar sound frequencies as humans * Historical analyses (e.g., of language evolution) * Tells us… * how words change meanings over time * how new words are come about * Why different communicative abilities evolve * what words are central to communication; what words are less relevant * Study: Compare Spanish in Europe vs South America * Looked at hominins (ambiguous words) – these words have different meaning based on the context they are in * Ex. river BANK vs money BANK * Results: * 50% variability in the language remains the same in two populations * 50% variability in the language - the meaning has changed * Reason: distinct cultural env, ext env, colonization in South America changed some meanings of words * Computational Modeling * Neural network simulation: dev an algorithmic or math model of how language works in the brain * Pros * Develop complex theory * Allows us to make predictions for empirical studies * Helps us to know what to study next * Allow us to do things that may be unethical on real subjects * Ex. damage a brain part in the model, and see how the model reacts to the damage * Ex. examine if therapy works for brain damage * Give different types of therapy to brain damaged model and see how it reacts to each one of it * Then test and apply this to humans, and see if it works * Genetics * Looks at how single and combination of genes leads to different amounts of NT, which in turn may impact our language processing * Ex. Diff types of dyslexia may linked to different collection genes; this can be related to whether the guy or gal contributed the genes

Answer 14

**_Classic examples[EL1]_** * Broca’s and Wernike’s Aphasias * In many ways, the seeds of the current interdisciplinary science of language research * Behavioural Manifestation: * Broca: meaning intact, but not the sentence structure * BA. 44 and 45 * Wernicke: no meaning, fluid speech (word salad) * BA 42 * Very different language impairments, linked to different brain impairments **_A basic map of some language systems_** * Note: Many areas serve more than one language relevant or non-relevant ability * Most of our cortex influences our language * LH is more specialized language, esp visual language * Visual inputs, auditory inputs, audiovisual integration * Audiovisual integration * Low level: how movement of lips correspond to different sounds * Higher level: read text on the screen (subtitles) interact with the dialogue the actors is speaking in the movie * Amodal semantic combination: how we map info from one modal to another * I pat a dog, it feels furry; and you can hear the word “fur” * Articulatory planning: plans how to move our mouth to produce specific sounds * Volition: decide what we want to produce *  Language processing and production is a very complex task, yet our brain has come to do things very efficiently and effectively * * * [EL1]Understand the difference between the two aphasias Know where the brain regions are located

Answer 15

**_Tools of the trade_** * Language processing is extremely rapid (ex. when you hear multiple words per second) and takes place across a distributed set of brain regions * These brain regions also subserve other abilities (e.g., auditory system is not only for language) * What types of methods can be used to probe language? **_Neuropsychology_** * We study patients with language impairments, link those to brain damage, usually originally primarily through post-mortem analyses * but sometimes we study living individuals like Phineas Gage). * Now, the brain is also explored through non-invasive techniques except in special circumstances (e.g., implant electrodes in pre-surgical epileptic patients’ brain). **_Behavioural Studies (aka Psycholinguistics)_**[EL1] * E.g., the McGurk effect * We use different information from the visual and auditory modality to perceive an ambiguous stimulus * Can provide important insights and constraints on cognitive and neural theories of processing * These types of tasks are often used in conjunction with other cognitive neuroscience techniques (e.g., ERP). * * * * [EL1]Understand how there are different systems contributing to language at the same time, some may be more powerful or relied upon more * McGurk effect is a great example of this

Answer 16

**_ERP_** * Behavioural studies are precursors of imaging studies * Pros: Excellent for studying fast processing, which is common in language (we need good temporal resolution; helps us track changes over time). * Cons: Relatively poor spatial resolution (without convergent evidence from other techniques/studies, data alone only tell you roughly which quadrant of the brain the signal came from). * Don’t know the precise location where the signal is coming from Example from Steinhauer, 2014, Applied Linguistics * Method: * 1. Put on EEG cap * 2. Listen to sentences * Ex. John ate broccoli at dinner * Ex. John ate democracy at dinner * Results * Blue vertical line: onset/when the words (i.e. broccoli or democracy) are spoken * Results: brain signals look similar overall, except at N400 (blue = difference) * N400: (400ms after onset) * There’s a difference b/w processing the word that makes sense vs one that doesn’t make sense * N400 relates to how comprehensible the sentence is

Answer 17

**_Example of Combining Behavioural and Neuroimaging Techniques_** * From Baart, Armstrong, Martin, Carreiras, and Frost (2017) * * Method[EL1] * Exp 1: behavioural experiment * 1. See and hear a Spanish word (ex. Apio) * 2. Press a button if what they see and hear match * Participants had to go through 4 conditions * no noise * audio noise * visual noise * audio and visual noise * Exp 2: same experiment but measure subject brain activity w/ EEG * Only difference: there is a 1250 ms b/w read + listen to word and pressing the button * Purpose: the actual act of pressing the button will activate the motor cortex * this delay allows researchers to differentiate b/w when the brain is reading + listening to the word vs pressing the button * Noise conditions * No noise: white text on black background * Visual noise: white dots on the “white text and black background” * * Audio noise * Ex. play a random audio noise -\> most ppl don’t know what word it is saying * When someone told me the word is “actress” * It seems obvious the word is actress in retrospect * This suggests that when the audio information is not clear, presenting the word visually can clear up the ambiguity * A + V noise (have audio and the visual noise presented) * Results – behavioral exp 1 * * Accuracy & RT: Ppl are the most accurate and fastest when there is no noise * Auditory noise makes ppl slightly less accurate and a bit slower * If you have more noise (i.e. A + V noise), you are the least accurate and the slowest * But the effect is not as bad as it seems * When we look at RT, AV noise is significantly slower compared to no noise * For accuracy, the drop is not as bad * Results = ERP Exp 2 * Researchers did a lot of t-tests * After controlling for this (the orange and blue), we get the results (black) * The effects happen around N400 (~500 ms) * This suggests that the integration of audio and visual noise is happening at the understanding/later semantic level (not early perceptual level) * **_Integrating insights_** * Combined data point to cross-modal integration occurring at a relatively late time-point, at the level of lexical/semantic processing * (combining info from 2 modalities happen later) * Surprisingly, the constraints from sublexical units (individual letters / sounds) seems relatively weak. (most integration happens at lexical semantic lv) * Potentially due to language tested (Spanish). * Transparent orthography: every letter maps onto a particular sound, and vv * Results may not apply to other languages that do not have transparent orthography (ex. Eng, Hebrew) * Points to value of Cross-linguistic comparisons another type of study that transcends any particular technique * * * * [EL1] Understand these experiments well and the significance of their results