Paul Steven Scotti

Userboxes

GWUThis user attends or attended
George Washington University

This user is a college student.

This user is a neuroscientist.

This user is interested in open access to research literature.

Anything written in block quote like this is typically a related example or a direct quote.

Your brain is like your own little universe. And they look basically the same! Coincidence???

Brain Anatomy

Cortex is brown. Corpus Callosum is white. Thalamus is red. Hypothalamus is salmon. Midbrain is blue. Pons is purple. Cerebellum is green. Medulla is yellow. Spinal cord is lime.

The central nervous system is composed of the forebrain (cortex, basal ganglia, thalamus, hypothalamus), midbrain, hindbrain (pons, cerebellum, medulla), and the spinal cord (not part of the brain). Note that the brain contains more than 100 billion neurons all working together in complex neural networks.

General brain subanatomy

Cortex (Telencephalon)

The cortex is like a large thin-crust pizza in terms of size and thickness (2 to 4 mm, which okay sure it's a really thin-crust pizza). To fit it in our heads it needs to be folded up a lot.

• Folded for increased surface area, folded sheets average 2.5 mm - gyri are the bulges, sulci are the grooves inward (the hollow part at the bottom is called fundus/fundi).
• Pial refers to the outside surface of the brain, as opposed to gray/white matter which represent inside
• Less than half a cm thick!
• Divided into 4 lobes, with 4 major divisions:

Longitudinal fissure divides left/right hemispheres

Central sulcus divides frontal/parietal

Lateral (aka Sylvian) fissure divides temporal from frontal/parietal

Parieto-occipital sulcus divides occipital from parietal/temporal

In terms of brain slicing:

• Horizontal slice is transverse or axial (parallel to floor)
• Shoulder to shoulder slice is coronal (like a tiara/crown)
• Nose to back slice is sagittal
• Looking at side outside: lateral
• Looking at side inside: medial (mid-sagittal typically)

In terms of anatomical location:

• Superior (aka dorsal) is toward scalp (opposite is inferior aka ventral)
• Anterior (aka rostral) is toward nose (opposite is posterior aka caudal)

Frontal Lobe

Involved with executive control

Prefrontal Cortex

Precentral gyrus (Motor cortex)

Parietal Lobe

Integrates sensory information from various modalities (particularly in regards to space). Contains somatosensory cortex. End location of dorsal (where) pathway in visual system. Parafoveal info and orientation. Can test this area using landmark test.

Landmark Task: Damage to posterior parietal causes impairment on landmark discrimination. Monkeys had to choose the covered foodwell closer to a tall cylinder (aka the landmark), which is positioned randomly.

Postcentral Gyrus aka S1 (or somatosensory cortex)

Area of sensory receptors, like temperature, tactile perception, pain, and other sensory modalities like proprioception.

Parietal-occipital Cortex

Involved with implicit memory (along with limbic system and striatum)

Inferior Parietal

Space-based attention is located in inferior parietal cortex - specifically intraparietal sulcus (find cite?).

Temporal Lobe

Medial temporal

Includes a system of anatomically related structures that are essential for declarative memory (LTM)

Auditory Cortex

Superior Temporal Gyrus

Wernicke's area: Involved with comprehension of language.

Object-based attention is involved in STG.

Transverse Temporal (Heschl's) Gyrus

Found in primary auditory cortex - first to process incoming auditory information.

Medial coronal slice showing medial temporal

Occipital Lobe

Involved with visual processing.

Calcarine sulcus/fissure

Calcarine sulcus

Where V1 is concentrated.

Basal ganglia (Telencephalon)

The basal ganglia are associated with a variety of functions including: control of voluntary motor movements, procedural learning, routine behaviors or "habits" such as bruxism, eye movements, cognition and emotion.

Striatum

The primary input to the basal ganglia system. A critical component to our reward system -  it receives glutamatergic and dopaminergic inputs. Is divided into dorsal striatum (caudate nucleus and putamen) and ventral striatum (nucleus accumbens and olfactory tubercle).

Spinal Cord

|Neuroanatomy}}

Processing

Bottom-up processing: stimulus-driven

Top-down processing: goal-directed information extraction

The brain is always active: https://www.youtube.com/watch?v=hnFCLlGeI98 (EEG), sometimes the activity of the brain shows it being restricted and other times it rapidly spreads across the brain almost enveloping the whole thing. Everyday activities showcase the brain being highly dynamic.

Hemispheric Lateralization

Left

• Language and speech processing occurs in LH for majority of people (Broca's and Wernicke's areas located on the LH)
• Visual Word Form Area (VWFA) is usually located on left fusiform gyrus.

Right

• Facial processing almost always on the right hemisphere (rFFA) -- perception of body posture and prosody are also lateralized similarly
• Odor perception is more lateralized to the right hemisphere - note that nasal passage to the brain is primarily ipsilateral
• The right hemisphere is dominant in emotional expression
• The right hemisphere is important for processing primary emotions such as fear while the left hemisphere is important for preprocessing social emotions.

Note that the processing of visual and auditory stimuli, spatial manipulation, facial perception, and artistic ability are represented bilaterally, but may show a right hemisphere superiority.

In left-handed people, the incidence of left-hemisphere language dominance has been reported as 73% and 61%, suggesting left handed people tend to be less lateralized than right-handed people.

Vision

Non-visual processing in V1

V1 is more than just a reception point for signals from optic nerve and thalamus. There is connectivity between V1 and A1.

Congenitally blind individuals had activity in V1 in response to spoken language, but this activity was missing in sighted individuals - showed evidence for neuroplasticity.

Could this be the strengthening of existing connections between A1 and V1? If V1 connections to non-visual inputs were disconnected would there be any symptoms? Note that V1 activity also appears with braille usage, but this is still functionally vision-specific.

Critical Periods (orientations)

Hubel and Wiesel 1970

We must feed our brains with experiences. In landmark studies with cats where lenses prevented them from seeing certain orientations (vertically or horizontally oriented features), their visual system was later incapable of seeing those orientations. The kittens would later run into chair legs as if they could not process corresponding visual information.

Gestalt principles

Rules and heuristics of the organization of perceptual scenes.

Figure-ground articulation

Although a fundamental aspect of field organization, it is not usually referred to as a Gestalt law. Usually the laws refer to somewhat more complex visual fields.

Why is the face-vase illusion hard to differentiate?

There are competing cues for figure-ground assignment. Convexity, familiarity, centeredness, symmetry, unlikeliness, and closure are all competing cues.

Neurons in V2 (second cortical processing stage) are selective for "border ownership". Meaning cells here are active when there is an "edge" in a region and the edge has a region on one side but not the other. Because the face and vase interpretations are both spatially consistent, the brain selects one, habituates, then switches to the other and back again.

Faces and Places

Facial processing is located in the rFFA (right fusiform face area) and place processing in the PPA (parahippocampal place area).

rFFA

The FFA is specialized for processing faces - see Kanwisher's research. Currently there are 2 views for how this area works:

1) Faces are "special": FFA is devoted specifically for recognizing faces.

2) FFA indicates area of "expertise": It's an area of perceptual expertise and faces are represented holistically like experts represent their objects of expertise. It's just the case that all humans are perceptual experts at face recognition because we naturally have a ton of practice.

PPA

The ventral temporal parahippocampal place area (PPA) has been implicated in scene processing, but scene information is contained in many visual areas.

Kravitz (2011): PPA representations were defined by expanse (open or closed scenes) and pEVC (peripheral early visual cortex) representations were defined by distance. PPA found to reflect spatial and not categorical or contextual aspects of real-world scenes.

Retinotopy

Retinotopy is the mapping of visual input from the retina to neurons (tonotopy is the equivalent for auditory). Retinal photoreceptors send information to LGN of thalamus which then goes to V1. There is many to one convergence, with many photoreceptors synapsing to one ganglian then to one LGN and then to a single V1 cell. Every cell is thus integrating more and more information. Adjacent neurons have receptive fields that include overlapping different portions of the visual field. Information becomes more and more rich as it goes from V1 to extrastriate regions.

pRF (population receptive field) mapping

You make a prediction of when a voxel will respond depending on when the stimulus image appears in that voxel's receptive field. Involves checking millions of receptive field possibilities in a x vs y vs sigma (radius) matrix. Each check produces a predicted t-series which is compared with the actual R^2 activity graph. The best similarities then tell us what each voxel's activity and receptive field is like.

Hierarchical model of vision

As we go down the visual processing stream, receptive fields grow larger and neurons integrate information from multiple areas.

Hierarchical model says we have a pyramid system:

1) Early vision: Oriented edge detection neurons, the first stop in the primary visual cortex

2) Mid-level vision: Motion, color, etc. Takes place in parietal lobe.

3) High-level vision: Recognition (e.g. faces in FFA). Takes place in temporal lobe.

Vision Search (Mitroff articles)

Mitroff Biggs Cain (2015): Multiple-Target Visual Search Errors: Overview and Implications for Airport Security: Multiple-target visual search errors refers to how a target is less likely to be detected if another target was already found in the same search. Raises 3 policy suggestions from visual search to airport security: remote screening away from the checkpoint (reduces anxiety, impatience, allows simultaneous screening); reduce items to lessen cognitive burden; emphasize search consistency in the training process. These suggestions can also apply to many search environments, not just airport security.

Mitroff (2014): What Can 1 Billion Trials Tell Us About Visual Search?: Discuss how 2 examples from Airport Scanner can address questions about visual search. 1. Ultra-rare Targets: Half all trials had at least one item of contraband, and ultra-rare means that the specific target appeared in less than 0.15% of all trials. Such rarity on visual search led to the finding that 'ultra-rare' items were much more likely to be missed than frequently occurring targets. 2. Multiple-Target Search: searchers with multiple-targets are much more likely to induce error such that if one target is found it is likely the other will go unnoticed. Named the "subsequent search misses" (SSM) phenomenon. Three primary theories of SSM have been proposed: (1) Original SSM explanation is that searchers become satisfied with the meaning of their search on locating a first target and terminate their search prematurely. (2) a resource depletion account suggests that cognitive resources like attention or working memory are consumed by a found target, leaving fewer resources left to find another target. (3) perceptual set account suggests that searchers become biased to look for additional targets similar to the first found target (finding a tumor and going into ''tumor mode", and becoming oblivious to bone fractures).

Biggs & Adamo & Mitroff (2015): Mo' money, mo' problems: Monetary motivation can exacerbate the attentional blink: Attentional blink (AB) is when observers are less likely to identify a second target when it appears approx 200 ms after a first target in a rapidly presented stream of items.

Attention

Can be either object-based or space-based. They are not mutually exclusive possibilities.

Space-based

Is like a spotlight (purely spatial representation of visual field) (inferior parietal lobe - as shown with neglect patients who only attend to right visual space)

Posner Cueing task: Cue above fixation of an arrow pointing left, right, or both. White target of fixation appears either to the left or right. ERP variant shows that attention modulates neural response despite physical stimulus staying the same, in which they are told to pay attention to either left or right side (still fixate center). More response to bar appearing in attended side.

Object-based

Involves preattentive segmentation of the visual scene (superior temporal gyrus)

Patients with neglect may copy images uniformly throughout visual field but only the right side of all objects - they have damage to STG.

Duncan (1984): Outline box with another line superimposed across is presented, followed by mask. The box and line both had two possible properties, and subjects--when asked to judge two properties of same object--were much better in performance than for discerning a property of both box and line. Shows attention is better on same object.

Cue was brightening at one of four rectangle ends. ISI = interstimulus interval (200ms) RTs less than 150 ms were excluded as anticipations or false-alarms.

Egly (1994): Wanted to make clear whether both space and object-based attention can apply in the same situation. They used a modified spatial precuing paradigm. To measure space-based, they cued to one location within object and examined performance difference for the cued part of that object versus uncued part of same object. To measure object-based, the target might appear in either same or different object while being equally spaced apart. Space-based evidence: Invalid cues take longer, even in same object. Obj-based evidence: Takes longer to switch attention to other object equidistant away. Also saw that parietal-damage patients showed greater cost for invalid cuing for contralesional targets (would later get specified to STG).

Exp 1-4

Exp 5

Shom and Yantis (2002): Is attention an early sensory representation, "spreading" within an attended obj and then stopping at borders? Or is attention a later object-specific prioritization strategy? Note that the compatibility effect (aka flanker effect) is studied: Responses are better if consistent to stimulus (eg if told to press left for blue and right for green, and blue is presented on the right side of the screen, RT will be lowered as it is incompatible). In this case, compatible response is when the flankers and the target would both necessitate the same button press in response. In experiments 1-4, subject was cued to attend to center of object, and flankers were mapped to either compatible or incompatible response, but equidistant from target. Note that flanker compatibility effect increases the closer the flankers are to target. Idea was that if attention spreads within object and stops at borders, then target and flankers appearing on same or different object would show different results, but it didn't! Thus, early sensory enhancement of attention does not respect object boundaries (space-based). So they looked at attentional prioritization idea - when attention is not focused in advance, object-based effects should emerge. Experiment 5 looked to see if this attentive state determined object-based effects. Attention was not focused in advance by the use of cue - an uninformative cue flashes briefly in central rectangle (assumedly summoning attention to central rectangle), but then the target can appear in either the cued rectangle or one of the two uncued rectangles. The compatibility effect was small with target in cued rectangle but large in uncued rectangle, because in this case the flankers are queried before the target. Shows that when attention is not focused (has to be deployed to multiple possible locations), object-based attentional prioritization operates, producing modulation of compatibility effect. Basically, if attention starts at one object, other locations in that object enjoy attentional advantage, even if target is not within that object.

Experiment paradigm, note the SOA

Shom and Yantis (2004): Looked at relative contribution of two factors to assignment of attentional priority. 1) Configuration (object-based) of objects in a scene--meaning object-based selection arises only when multiple locations in the scene must be attended, and 2) Contextual (probability) prioritization, the probability of the target appearing in each location imposed (i.e. cue validity). They varied the time between cue and target being shown (SOA - stimulus onset asynchrony) to determine whether relative contribution of these factors differs over time. Results found that at short SOAs, both factors influenced performance, but at longer SOAs, the configural contribution disappeared and RTs depended exclusively on contextual (probability) contribution. Thus, configural contribution is fast-acting and seems to be a default mode of attentional prioritization - whenever there is more than one object in a scene, regions within attended object are assigned higher priority for visual exploration. But given hints to contextual contribution and a few hundred milliseconds, and this effect wears off.

exp 1 stimuli

Richard, Lee, Vecera (2008): Focused on attentional spreading (early sensory representation refuted in shom 2002) by using a flanker task where the target was fixed and known to participants. A spreading attention account predicted that object-based attention will rise from the spread of attention through an attended object (otherwise there would be an object-based effect and a prioritization account would be concluded, but it wasn't). Participants had to say if center bite was circular or rectangular (given either compatible or incompatible flankers). Even though participants knew that the target would appear at fixation, the results still demonstrated larger flanker effect when in single object. Shows that spreading-like effects are observed when task-relevant features are part of an object and contribute to the shape of the object - we can't help but have our attention spread throughout it!

Drummond and Shomstein (2013): With high spatial uncertainty, all information within the stimulus display is used for attentional selection. As certainty increases, however, only spatial certainty guides selection (object ends and whole objects are filtered out) (contextual probability all that matters).

Shom and Johnson (2013): Effects of Reward on Space and Object-Based Selection. Found that reward alone, not the objects, guides attentional selection and thus entirely predicts behavior - this suggests that guidance of selective attention, although automatic, can be adjusted in accordance with external nonsensory reward-based factors. Essentially it's the Egly experiment with greater reward for answering in the different rectangle (and in the 2nd experiment, rewards of 1 to 6 points were randomly given on both different and same objects).

Note: When summarizing all of this look at Shom's 2nd to last opinion article.

Auditory and Language

Speech

Aphasia

Aphasia is the loss of ability to understand or express speech (tan tan tan).

Broca's aphasia: Inability in speech production

Wernicke's aphasia: Involved with comprehension of language. Aphasia here involves words making no sense and no syntax, but would sound fine to someone who does not comprehend the language the patient is trying to speak.

Arcuate fasciculus – transmission channel from Wernicke’s to Broca’s area 

Hearing

Right ear bias to perceptual discrimination

If you want to whisper to someone and have them comprehend most clearly, whisper into their right ear.

Language

Critical Periods (learning new language)

Perani and Abutalebi (2005)

Much easier to learn another language if one is exposed to it as a child than if one waits until adolescents or adulthood. Research suggests that languages learned during childhood are organized in the brain similarly to native language, while there is a different somewhat more superficial organization of nonnative languages in the brain.

Olfactory

Odors smell better through right nostril, odors named more correctly through left nostril

linguistic processing is primarily a left hemisphere activity, whereas recent evidence has shown that basic odor perception is more lateralized to the right hemisphere. Importantly, under certain conditions, emotional responding also shows right hemisphere laterality. Hedonic (pleasantness) assessments constitute basic level emotional responses. Given that olfaction is predominantly ipsilateral in function, it was hypothesized that odor pleasantness evaluations may be accentuated by right nostril perception and that odor naming would be superior with left nostril perception.

The study's results backed up their hypothesis. I imagine the idea that odors "smell better" through the right nostril also means that horrid smells would smell worse through the right nostril, similarly activating more emotional responses.

Memory

"You have to begin to lose your memory, if only in bits and pieces, to realize that memory is what makes our lives. Life without memory is no life at all ... our memory is our coherence, our reason, our feeling, even our action. Without it, we are nothing." - Bunuel Portoles, 1983

LTM

Declarative (explicit) Memory

Remembering particular events, or facts and figures.

Episodic Memory

Person's unique memory of a specific event.

Semantic Memory

General knowledge of the world accumulated throughout our lives.

Implicit (nondeclarative) Memory

Memory in which previous experiences aid performance w/o conscious awareness. Generally categorized as either procedural memory or priming.

Procedural Memory

Remembering how do do things, like ride a bike, sing a song, drive etc.

Perceptual Priming

Implicit memory effect: exposure to one stimulus influences response to another. If given a list of words to read, with one of the words being "table" and then asked to think of a word starting with "tab" they are more likely to think of "table".

Classical Conditioning

Form of learning where you learn that one stimulus predicts that an event will occur.

Nonassociative Learning

Aka habituation - form of learning where you learn to stop responding to a stimulus that is no longer relevant (not getting shocked by loud noises if they are repeatedly broadcasted).

STM

Patient HM: Had epilepsy and bilateral medial temporal lobectomy (inc half of hippocampus) - all seemed fine but he couldn’t form new long term memory, but implicit memory was intact. Threw wrench in idea that ltm and stm are a monolithic idea.

Reconsolidation Hypothesis

Holds that when a memory is recalled, its molecular trace in the brain becomes plastic. So every time you pull up a memory it is effectively erased and saved again - your memory of an event literally changes every time it is brought up.

A drug that blocks memory formation would therefore not just block new memories but also block any reactivated memories. Could this wipe memories? Gisquiet-Verrier et al. proposed that amnestic drugs would instead add an additional element to reactivated memory trace - the memory of the amnestic itself. The drug tags memories with amnestic-intoxication which makes them later less accessible because of state dependent recall - they could only be easily retrieved by taking another dose of amnestic. Tested the theory on rats given cycloheximide and were supported.

Alzheimer's Disease

AD is the world's leading form of dementia. Currently affecting about 47 million people worldwide it is expected to double every 20 years, noticeably in the developing world.

Nerve Growth Factor (NGF)

Preliminary findings from the first human trials designed to test potential benefits of NGF look promising.

Transactive Memory

The internet can give us the illusion of knowledge, thinking we are smarter than we really are. This is because we confuse our personal knowledge with knowledge we have access to. If you have ever left something you need for work by the door the night before, you've used transactive memory.

Asked people to provide answers to factual questions - half were instructed to look up the answers on the internet before answering. People who used the internet were significantly more confident about their understanding of following sets of questions. They really thought the knowledge was theirs - the confidence effect persisted when the control group were provided answer material and the internet-search group were instructed to search for a site containing the same material. Something about actively searching for the info led to the illusion.

Sleep

Sleep position

Sleeping on your side drains CSF better than lying on your back or side, so it is more effective for your brain's self-cleaning process. It also helps with snoring, as otherwise there is the possibility of the base of the tongue collapsing to the back wall of the throat, which causes a vibrating sound during sleep.

REM sleep behavior disorder (RBD)

Patients make movements, sometimes violent ones, while they sleep. Sometimes these movements are accompanied by speech. Their movements seem to correspond to the content of the dreams that the patients are experiencing.

Notably, some RBD patients report never dreaming. Howeverin, in many cases their actions during sleep suggested they were dreaming. It is thought that all people dream, and those that say they never have dreams just can't remember them (not necessarily proven yet though). This introduces a difficult problem - how do we ever find out if there really is a genuine non-dreamer out there?

Plasticity

Critical periods demonstrate the influence of early experience on organization of certain functions of the brain, but the brain is not static and is constantly changing. It's a very malleable organ.

Initial evidence

Merzenich et al. 1984

Monkeys that lose a finger reorganize the brain so that the area corresponding to the missing finger begin to respond to stimulation of adjacent fingers. Suggests mature brain is capable of reorganization and is more malleable than previously thought.

Brain Scanning

To reveal the structure or function of the brain we use scans or lesions. Scans can either be structural (CAT, MRI) or functional (fMRI, PET). Structural looks at the anatomy of the brain while functional looks at how the brain is functioning.

CAT

CT or CAT (computerized axial tomography) are structural imaging scans which are glorified x-rays from all directions. One reconstructs differential absorption of rays through how bones absorb a lot, blood a little, and neurons absorb somewhat.

To compare CAT to MRI:

• CT works well for imaging bone structures.
• Some patients who have received certain types of surgical clips, metallic fragments, cardiac monitors or pacemakers cannot receive an MRI but can have a CT scan.
• The time taken for total testing is shorter in CT than for MRI.
• CT can be more comfortable for patients who are claustrophobic.
• MRI uses magnetic field while CAT uses radioactive x-rays
• MRI has much higher detail for tissue structure
• MRI requires patient to stay very still, which is not as much a crucial thing in CAT scans

MRI

Great video on understanding how MRI (and also fMRI) works: https://www.youtube.com/watch?v=MiL0wCZr0Mw

MRI’s structural imaging views at a high resolution the difference between tissue types by studying water molecule precession (look at water density, perturb all of them, look at water density again). On the other hand, an fMRI’s functional imaging views the tissue differences with respect to time. In addition to this, an MRI has a higher spatial resolution than fMRI, but a worse temporal resolution. Even so, the temporal resolution of fMRI isn't that great either, with there being around a 6 second delay between neuronal activation and seeing any lighting up from the scanner.

Note that scanning time, image resolution, and signal-to-noise are all related factors that interfere with data.

DTI

Diffusion-tensor imaging (DTI) is an MRI technique that can measure macroscopic axonal organization (aka look at the tract pathways). Instead of lighting up brain areas like fMRI, DTI shows white matter pathways. Uses idea of anisotropic diffusion (metaphor to understand: drop of ink on a wet paper and it moves in a certain direction) that suggests higher density of fibers oriented in a direction - used to estimate organization bc water more easily moves along these axonal bundles. Only thing is we don't know anything about directionality. The water diffusion is measured along the gradient axis - meaning out of the x y z we pick the best looking axis for data analysis.

VBM

Voxel-based morphometry (VBM) involves a voxel-wide comparison of local concentrations of gray matter between subjects. After first segmentation (identifying white and gray matter through MRI), VBM registers every brain to a template (a process called normalization, and the template is called a deformation field). This process allows us to discount global shape difference between individuals. Then the brain images are smoothed so that each voxel represents the average of itself and its neighbors. Finally, images are compared across brains at every voxel.

Normalization is tricky and easy to fudge because it depends on good alignment.

Currently each voxel is assigned a probability of belonging to either white or gray matter based only on intensity and probability of neighbors (if surrounded by gray matter it is likely also gray matter).

fMRI

To briefly summarize: Increased neural activity leads to increased blood flow, which changes the relative concentration of oxygenated and deoxygenated hemoglobin - these two things are magnetically different and the fMRI picks this up.

What to consider when evaluating fMRI articles:

1. Whenever you see an article that's like, "Region X is activated when subjects do Y", always ask "in comparison to what?" This is because the brain is always active - fMRI only makes sense through differences between conditions.
2. Specificity: If brain activity goes on in a region that is known to be involved with process Z, that doesn't necessarily mean whatever activity prompted that region to be activated is provoking that process. In other words, saying that your task provoked brain area X to light up, and a previous study showed that brain area X is involved with Y doesn't mean that same task is necessarily involved with Y. This is reverse inference. That region could be involved with many different processes. Activity in a certain region doesn't mean shit unless you cite or prove that there is a very strong case to be made that activity in this region always provokes a specific process. E.g. Kanwisher had to prove through like 20 alternative hypotheses that FFA is only specific to facial processing. Shown in the highly cited 'empathy activates physical pain regions' paper.
3. The experiment needs to be replicable - they need to lay out the methodology to enough extend you could replicate it and if you can't, it's impossible to really trust the data.
4. Different peoples brains are physiologically different. Are the brain regions activated for these subjects really activating the same corresponding places? Alignment can be finessed to "look" better. The only way to get past this is through 'within subjects' design, or comparing tasks in the same individual, which removes that ambiguity of what counts as the same place. Although there is still some problems with a subject moving in the scanner.
5. Statistics are not transitive. If Group A (normal people) shows significant differences between tasks and Group B (autistic) doesn't, you need to compare them directly statistically - you can't just conclude from one group showing significance that Group B participants process differently (because it could be the case that there are huge error bars they didn't consider if you compare them). Even when it's specific to a brain area, saying they found significance in FFA for X and not Y and then saying therefore FFA responds more to X than Y - that's not true, you need to still statistically compare them.
6. A voxel could incorporate a million neurons and those neurons could be involved in different processes, creating ambiguity in the role of the voxel.

PET

Positron emission tomography (PET) scan measures blood flow in the brain by injecting a radioactive isotope which will quickly decay in the blood. Blood flow increases in areas of brain usage so one can see those isotopes in the scan. Test is usually run twice - once as the control and once as the experimental. The difference is measured and compared. Similar to fMRI in that both measure blood flow, an indirect measure of brain activity. Almost always used in clinical rather than research settings.

In comparison to fMRI:

• PET does not require very still subjects
• Spatial resolution is lower in PET scans (5-10 cubic mm versus 3 or less cubic mm in fMRI)
• There are more hospitals that have MRI scanners than PET scanners, making it less convenient
• Requires radioactive isotopes which can only be given a few times before it's unsafe
• Higher cost for PET scanners

ERP (from EEG)

Event-related potentials (ERP) - methodology drawn from EEG (electroencephalography) which simply records electrical activity that is happening over the brain. Electrodes pick up neural communications (aka action potentions) between the area immediately underneath the electrode on the scalp and a reference electrode placed at an inert spot like an earlobe. It reflects the the product of perhaps millions of neurons - it's like looking at a satellite image of the earth, showing general layout but not much detail. Has great temporal resolution but very poor spatial resolution. ERP averages multiple EEG segments together and forms an averaged ERP waveform. A huge drawback to ERP is that we cannot know where the signal is coming from.

MEG

Magnetoencephalography works by recording weak magnetic fields and averaging many trials. Has low spatial resolution but great temporal resolution. The MEG equivalent of ERP is the ERF, or event-related field

TMS

Transcranial magnetic stimulation is an induced lesion method that is reversible and not invasive. Researchers apply an electrical stimulus directly to the brain without having to open the skull surgically and install electrodes. Instead, the stimulus is applied by a magnetic field that passes through the skull directly and painlessly. It is very important to position the TMS coil precisely above the brain region to be imaged. In order to do so, a magnetic resonance image of the person’s brain is often taken first, so that the coil can be adjusted for its particular pattern of folds. Can only affect the outer surface of the cortex (poor spatial resolution) but has good temporal resolution.

Methodology and Analysis

fMRI

Event related vs Block related

efMRI can be used to detect changes in BOLD to neural activity. Currently there are two ways that are typically used to present stimuli in fMRI designs - block related design (aka boxcar design) and event-related designs. Block-related involves two or more conditions alternated in order to determine difference between conditions. Event-related designs are not presented in a set sequence - the presentation is randomized and the time in between stimuli can vary. In block related designs and non-rapid event-related fMRI, long ISIs are used. This is potentially bad because mental operations being performed by a subject during any given task become increasingly unclear as the amount of time in between adjacent stimuli is increased (e.g. consider any memory based tasks).

Because of this, only use event-related averages if used in rapid fMRI paradigms when a sufficient degree of randomization can be introduced into the stimulus presentation order. Otherwise use block.

Event-related designs are single stimulus, single response designs. This leads to less power, and less able to detect differences, so you have to run many conditions -- but it leads to very flexible data analysis where you can aggregate data in different ways or eliminate unnecessary conditions. It's difficult to look at a subset of a block individually in block designs.

Deconvolution

In typical efMRI, after every trial the hemodynamic response is allowed to return to baseline, but that takes like 6 seconds. In rapid event-related fMRI, there are two ways to estimate BOLD timeseries: event-related averaging and deconvolution.

Serences 2003:

Event-related averaging: Every possible combination of trial sequences must be used and the ISI intervals jittered to extract the TRs effectively for each individual stimulus. In fixed ISI you can’t easily decouple the onset times for different combinations of overlapping responses. If there is variability, like in jittered ISI design, you can more easily tell the offset times because they are no longer perfectly correlated. It's not always best because variance in stimulus timing may complicate interpretation of data of underlying cognitive operations.

Deconvolution (using Ordinary Least Squares estimate):Doesn’t need jitter, but it assumes that each response follows a certain pattern (it's not able to detect responses which don’t look a certain way) - usually a good trade-off. The only responses that don’t look like normal are usually subcortical and don’t involve vision. Most experiments therefore assume the shape of the hemodynamic response and avoid this problem. Having restricted ISI distributions often maximizes the ability to make psychologically valid experimental conclusions.

Multidimensional Scaling (MDS)

MDS used in Kravitz 2011

A means of visualizing the level of similarity of individual cases of a dataset. First generate ROIs (regions of interest) from significance maps by taking contiguous clusters of voxels that exceeded threshold and occupied the appropriate anatomical location. Then use MDS to position similarity matrixes on a 2D plane. Used when you have a large number of conditions and it produces data that can be directly correlated with behavior at an individual item level -- use it with event-related designs because those are on the level of individual stimuli.

Multi-Voxel Pattern Analysis (MVPA)

Better than univariate analysis - which involves taking all your voxel activations, averaging it all for a region, and comparing that region's average to the same region's average for a different condition of your experiment. In MVPA, you take all these voxel activations and compare each voxel individuals between regions in conditions. MVPA correlates all voxels to each other, so alignment can be more of a problem (need to isolate voxels between conditions), but it gives you better data. Both suffer from limitations of its correlative method of measuring activation, the limitations of voxels, and also the limits of fMRI.

AFNI

Used for analysis and display of fMRI data. Founded on idea that it controls for interactive adjustment of stat thresholds and colorization, lets one see intermediate results so it's easy to backtrack to understand how results were obtained, and everything is open-source.

(Dynamic) Functional Connectivity (DFC)

Functional connectivity changes over a short time when observed, but we presume that functional networks are actually static. Usually compares specific activity against dynamic mode network (DMN), which is a network of brain regions that are active when individual is not focused but at wakeful rest. Many argue that DFC is a simple reflection of analysis, scanner, or physiological noise.

Involves linear relationships in the neural network, when in actuality networks are not linear (e.g. may find network from V1 to ITC, but there's another one going V1-V2-V3-V4-ITC). May produce interesting consistencies, like the DMN, but there needs to always be follow-up experiments. It's a quick way to get info out of a subject - doesn't require a design, code, and data can be related to other data to be more interesting. Needs converging additional experiments.

Functional connectome fingerprinting (2015)

Derived from Human Connectome Project - divided 128 subjects into 268 nodes (parcellation). Then constructed correlation matrices which encode the connectivity map (aka connectome) for each participant, which were used in a permutation based decoding procedure to determine how accurately a participant's connectivity pattern could be identified from the rest. Could decode identity from one resting state scan to another with high accuracy (over 90%), and could also do rest-to-test and test-to-test decoding.

Also did crucial alternate hypotheses: Head motion strongly influences functional connectivity, so they applied the same decoding approach to an expanded set of motion parameters. Motion was unable to predict identity, with less than 5% accuracy. 2nd alternate hypothesis was that accuracy could be primarily derived by anatomical rather than functional features, so they tested if classification accuracy decreased as data were increasingly smoothed (which would eliminate anatomical feature contribution), and this did seem to decrease accuracy.

Showed that while activity alone can make an individual unique, its the connectivity data that is needed to construct a complete fingerprint. Frontal-parietal and default mode networks showed highest contribution to fingerprint.

Showed that resting state network is highly stable and one can identify one's neural fingerprint. Didn't delve into vasculature or breath-related compounds.

P-Hacking

A talk from Neuroskeptic about P-hacking: https://www.youtube.com/watch?v=A0vEGuOMTyA

P-hacking is the use of data mining to uncover patterns in data that can be presented as statistically significant, without first devising a specific hypothesis as to the underlying causality.

1. “Overhacking”: researchers might not stop with a significant result, but might keep hacking their p-values to be as low as possible, to make their results look more compelling. There is no reason why p-hacking has to stop once p < 0.05. Indeed, as awareness of p-hacking grows, researchers may become increasingly suspicious of marginally significant p-values, and the de facto desired significance threshold may shift downwards. Perhaps this is already happening.

2. Selection bias: we tend to prefer the lowest p-value given a choice. If you find two results (or two variants of the same result), with p-values of (say) 0.04 and 0.07, it would be easy to report only the 0.04. But if you have three options, 0.01, 0.04 and 0.07, you’d probably report 0.01, not 0.01 and 0.04. In other words the pressure is not “publish all p-values below 0.05″ but “publish the lowest possible p-value, if this is below 0.05″. Importantly, this doesn’t even require ‘active’ p-hacking i.e. trying different variants of the same analysis, it just requires the running of multiple analyses (maybe on different variables) and selective publication.

3. Selective debugging: Sometimes researchers use inappropriate statistical tests, or there are artifacts or data coding errors. Researchers may be more likely to spot and fix ‘bugs’ that create non-significant p-values than those that produce significant ones. If an experiment ‘doesn’t work’, I look at my analysis procedures. If I spot a bug, I fix it, and run the analysis again. And so on. Once I do get a significant result, it is very tempting to stop looking for more bugs. What this means is that I’m selecting in favor of bugs that produce false positives. I consider selective debugging a form of p-hacking. But bugs can’t be assumed to produce evenly distributed p-values. Some bugs produce p-values clustered around 0, because there ‘really’ is a deviation from the the null hypothesis – albeit because of a bug, not because of the true data.

Ways to stop it:

"Two-Step" Review

Reviewer is not swayed by whether results agree or disagree with their own views, or whether results are positive or negative. Peer review occurs with just intro and methods. Then a secondary review for results and conclusions afterwards. Then publication.

Peer Pre-Review

Researcher writes intro and methods before doing the study. Then it gets sent off for review - journal commits to publishing paper whatever results are, even if null. Researchers have thus written protocol on record beforehand, so they can't change protocol to create more significant results.

Drugs and Chemicals

Nootropics

People who have a disorder that negatively affects cognitive ability have the potential to see a strong increase in ability because of medication (schizo, alzheimers, adhd). For healthy people, it's a much smaller effect and there is still much research needed to investigate the health effects of cognitive enhancement. Unfortunately not many studies are interested in nootropics so there is not much info about long-term use of any cognitive enhancers on a normal population.

I remember experimenting with nootropics in high school for several months - particularly piracetam, choline bitartrate, and l-theanine I believe. I never felt any dramatic effects and the effects I did experience I believe may have been placebo - things like feeling like I can socialize perhaps a bit better and feeling a bit more "awake". I eventually stopped because I didn't experience many noticeable effects, I felt anxious about my parents finding out, parachuting weighted drugs in the morning made me feel like a drug dealer, and sometimes my eyes would twitch in the mornings and I was worried maybe this was a side-effect.

Focus

The fuzzy-headed feeling from cognitive fatigue can be staved off through drugs used for increasing focus, like adderall or modafinil (reduce distractability or increase endurance). These drugs may be harmful though. Adderall can become addicting and may lead to severe anxiety and heart problems. Modafinil can also be addicting and if used for an extended time the body will build up tolerance. It's worrisome to think that modafinil, which has a relatively low number of side-effects, that people will take the drug just to "get ahead" without there being any hard scientific backing as to the risks of extended use.

IQ

IQ is more about the hard-wiring of the brain, largely determined by genetics

Oxytocin

Produced in hypothalamus and involved with intimacy, sexual reproduction, and social bonding. Released in large quantities during childbirth and helps with birth, maternal bonding, and lactation. Can be used to start or increase speed of labor and stop bleeding following delivery.

On trust

Belgian researchers published evidence that said oxytocin promotes trust in humans - but now they replicated their own experiment which failed to support their claim. Why did this happen? The original study was single-blind - the participants didn't know whether they were getting oxytocin or placebo but the experimenter did know. This may have influenced how comfortable the participants felt. It was actually notably brave that they replicated their own results in the first place, and it seems it was worth it.

Statistics

http://www.muhlenberg.edu/pdf/main/academics/psychology/stats_decision.pdf

http://www.graphpad.com/support/faqid/1790/

http://www.csun.edu/~amarenco/Fcs%20682/When%20to%20use%20what%20test.pdf

Differences of Groups

T-Test

Compares two groups on some variable of interest.

Do males and females differ in amount of hours they spend shopping in a month?

The datasets usually follow normal distribution curve, but variance is unknown and assumed to be equal. Better suited for smaller groups of data (if larger, use Z-test).

Chi Square

Compares observed frequencies to expected frequencies.

Is the distribution of sex and voting behavior due to chance or is there a difference between sexes on voting behavior?

ANOVA

Stands for analysis of variance.

Tests the significance of group differences between two or more groups. IV has two or more categories. Only determines that there is difference between groups, not which is different.

Do SAT scores differ for low, middle, and high-income students?

Two-Way Anova

Allows you to ANOVA with access to two factors simultaneously. Allows you to see the interaction between the parameters.

Do SAT scores differ for income level (low- middle- and high-income) and gender?

Interesting Psychological Experiments

Monster Study

Stuttering experiment performed on 22 orphan children. Gave speech therapy, half were praised for fluency and other half were belittled for every speech imperfection and telling them they were stutterers. Those in negative therapy experienced negative psychological effects (trouble freely expressing themselves, growing withdrawn, worsened schoolwork, etc.) and some retained speech problems the rest of their lives. One orphan who grew withdrawn said that she stopped speaking to her best friend and ran away from the orphanage.

Baba Shiv (1999) Affect and Cognition in Consumer Decisions

Participants were asked to memorize digits between two and seven digits and then walk down a hallway to report the numbers. But they get interrupted midway and told that in thanks for participating they can either have a chocolate cake or a fruit salad snack. The longer the digits they were to remember, the more likely they were to go for the cake. Thought that because availability of processing resources is low, spontaneously evoked affective reactions have greater impact on choice than cognitive reactions (debating the consequences of the decision).

Stephen Want's Attention and Thin Model Experiment

Participants either had to memorize an easy or difficult string of numbers, and then look through a series of images and recite the numbers at the end. The images were of very thin fashion models. The easy group were tested after the experiment and showed increased mood and appearance dissatisfaction, while the hard group showed no ill effect. The hard group was not able to do non-automatic tasks because they were preoccupied with remembering their number, which seems to suggest that comparing ourselves to thin-ideal images is a non-automatic process. This means that if we do not pay attention to the media's portrayal of thin as ideal, we won't be influenced by it, which is an optimistic thought.

The Candle Problem by Karl Duncker and how it sheds light on motivation

Given a matchbook, a tin of thumbtacks, and a candle, the task is to prop the candle so that when lit, no wax will fall down onto the floor or table below. It questions the idea of functional fixedness, in that one needs to realize that you need to use the receptacle of the tacks in order to mount the candle on the wall - it serves more than its apparent function of holding the tacks. Elaborated upon, there were two groups told to do the task. One group was given a financial incentive that the quicker they completed it, the more money they would get. This group took on average over 3 minutes longer despite given a greater incentive of physical reward. The other group was given no rewards and told it was just to average participant norms. Questions the idea that if you want people to work better that you need to reward or incentivize them. Extrinsic motivation or reward-and-punishment approaches only work for narrowly focused and less conceptual or creative tasks. For instance, take the tacks out of the box already for the participants and now the incentivized group kicks the other groups ass.

Also Dan Ariely's study as mentioned in the Ted talk by Dan Pink, larger rewards lead to worsened performance for any tasks that required even rudimentary cognitive skill. So how do we get intrinsic motivation for all these other tasks? We need autonomy, mastery, and purpose (to do something in service of something larger than ourselves).

Baby photos in wallets dramatically increase odds of it being returned

Richard Wiseman led an experiment where they dropped wallets on the sidewalk and observed whether people returned them or not. The control wallets with no photos only got returned 15 percent of the time, while the wallets with baby photos in them got returned 88 percent of the time (they used other photos as well but baby photos were by far the most significant). Note that I have personal experience confirming this study - I always carry a picture of me from elementary school (obviously I was an adorable little tyke) and I did have my wallet returned!

Michelangelo phenomenon (and Blueberry phenomenon)

Couples can mold one another both psychologically and physically - they begin to look more similar as resonant emotions sculpt their facial muscles into similar patterns - the Michelangelo phenomenon refers to this sculpting of each other mentally and physically. Opposite is the Blueberry phenomenon, in which interdependent individuals bring out the worst qualities in each other. (I wonder if they begin to look more alike as well, in a bad way?)

Babies remain calm twice as long when listening to song compared to speech

Found that babies listening to a novel song remained calm twice as long compared to speech. They seem to get "carried away by the music".

Rhythm 0 (1974)

Six hour performance art, involved artist standing still while the audience was invited to do to her whatever they wished, using one of 72 objects she had placed on a table (inc. rose, wine, a scalpel, nails, and a gun loaded with one bullet.). Was like a crazy zimbardo experiment - things started out peacefully, giving her a kiss and offering her a rose. By the end, all her clothes were cut off, she was cut up in places, and one person put a gun to her head and positioned her finger on the trigger and a fight broke out around her. She was willing to sacrifice her life. When she started moving after 6 hours, people couldn't confront her and physically ran away from her.

Meta/Other

Trends/Things to Keep in Mind in Neuroscience

Better measures -> briefer exposures

As precision with which the brain can be measured increases, we find briefer exposure to experiences produces measurable changes in the organization/function of the brain.

Effect of time pressure versus accuracy for subjects

When subjects are cued to emphasize speed, there is increased activation of the supplementary motor area and striatum (Forstmann et al 2008). When cued to emphasize accuracy, there is less activation within these same brain regions suggesting greater involvement of the inhibitory mechanisms associated with increased monitoring and control of ongoing actions.

Neuroscientists interpret studies differently and use loose jargon

http://fixingpsychology.blogspot.com/2012/12/neuroskeptic-part-1-misunderstanding.html

Misunderstandings occur often in neuroscience because people interpret the results literally, without understanding them in the context of professional norms and publication pressures. We are more likely to get published by using cool-looking jargon.

Low activation in face-processing area confirms deficits in people with disorder X. 

The public would interpret this in 3 ways neuroscientists would not:

1. Thinking there is a "face area" when in actually its more of a visual expertise area, and we know that countless other brain areas are crucial for face recognition as well (good chunk of visual cortex).
2. Thinking an intervention aimed at brain area could fix facial recognition problems.
3. Assumes that disorder X is real. Just because you have group differences doesn't mean that the differences are because of the disorder. Selectively grouping participants based on behavioral differences already can lead to group differences.

Consciousness

The problem with studying consciousness is that it cannot be experienced, it is experience. Just as "life" cannot be physically located in the human body, consciousness cannot be specifically located in the brain.

Ramon y Cajal

Spanish anatomist known as father of modern neuroscience. He won the 1906 Nobel Prize for discovering neurons and intuiting the form and function of synapses.

He described neurons as mysterious butterflies of the soul, the beating of whose wings might one day, who knows, reveal the secrets of mental life.

Phil Kennedy

He traveled to Belize to implant a large electrode in his motor cortex. Is trying to build a speech decoder. Was a very dangerous procedure and cost him a lot of time and money but wow.

Oliver Sacks

Sacks wrote what he called ‘romantic science’. Not romantic in the sense of romantic love, but romantic in the sense of the romantic poets, who used narrative to describe the subtitles of human nature, often in contrast to the enlightenment values of quantification and rationalism.

In this light, romantic science would seem to be a contradiction, but Sacks used narrative and science not as opponents, but as complementary partners to illustrate new forms of human nature that many found hard to see: in people with brain injury, in alterations or differences in experience and behaviour, or in seemingly minor changes in perception that had striking implications.

Listen to Oliver Sacks talk on RadioLab. He was such an entertaining guest and lived such an interesting life. If I ever find the time, I want to read his final autobiographical book.

Reproducibility Project

Reproduced 100 experimental and correlational studies in three psychology journals and found half the magnitude of original effects. 97% of original studies had significant effects, but only 36% of replications had significant results.

Neurosophisms

Coming from the word "sophisma" meaning "clever device", neurosophism is the fallacious application of neuroscientific language to make something sound more intelligent or complex, and it seems to be plaguing tons of faculties. For instance it is becoming increasingly prevalent in education, with sentences like this:

You can’t think when you’re stressed, you can’t learn when you’re anxious and that’s one of the primary principles of the neuroscience …

This implies that prior to neuroscience, teachers were unaware that stress and anxiety would negatively affect learning. People are repackaging information that would sound fine without any reference to the brain in order to sound more sophisticated.

----

current experiment

The question is - oba has been investigated in very simple objects, is it the case that attention benefits from items appearing on the same object? Bc it’s a simple boundary it constrains allocation of attention - but what if the object (weak object representation) is a real object, will you see still the oba or would it be much larger than with simple rectangle. Or maybe the OBA disappears because the two object representations maximally activate?

Fund question: Does the strength of object representation modulate object-based effects?

References

Note that most stuff here is not cited, and some just has an author in place of any citation.