Cognitive Science: An Introduction/Neuroscience Methods

What the Brain is Like
Although it only accounts for less than two percent of our body's mass, the human brain uses twenty percent of its energy budget--about three bananas worth of energy every day. But even this isn't all that much. This is a real energy limit that means that at any given time only one percent of your neurons can be active at a time.

Using Neuroscience to Help us Understand the Mind
Looking at how the brain works is an important way of helping to understand the mind. Whereas psychology often uses behavioral experiments to understand the mind, neuroscience offers several ways to look at the brain in ways that constrain our theories. Brain science has contributed to our understanding of the mind in many ways, including differentiating short-term and long term memory, episodic versus semantic memory, and implicit versus explicit memory.

Imaging
Many brain studies examine the activity of neurons. This is usually done with imaging. Most brain imaging shows brain consumption--either of oxygen or of glucose. The assumption is that consumption rates indicate information processing in the brain.

Magnetic Resonance Imaging (MRI)
Many atoms in the human body are hydrogen atoms. The single proton in each of these atoms acts like a little compass needle. MRI produces a magnetic field. Radio waves knock the protons out of alignment, and when they settle back down to point in the right direction again they release a radio wave signal that can be measured outside the body. This wave can be interpreted to figure out the three-dimensional structure of the tissue in a body or brain, because the atoms in different tissues align at different speeds.

MRI is used to describe the physical structure of the brain (or other tissues), not its processing over time. However, it can see the connectivity between brain structures. Using a technique called diffusion-weighted imaging, scientists can track the direction of water molecules to see nerve fiber direction and get an idea of how a brain's network is set up.

Functional Magnetic Resonance Imaging (fMRI)
Around 1990 Seiji Ogawa and colleagues invented a way to study the brain's activity in action without opening the skull or using an injection.

Active cells, wherever they are in the body, use up more oxygen in the blood. Blood rich in oxygen has a different magnetic properties than blood depleted of oxygen, so with a powerful magnet, you can observe which cells are working hard and which ones are working less hard. The fMRI machine does this. The researcher typically asks someone to do some task, or look at some stimuli, and the fMRI records brain activity, under the assumption that the part of the brain working the hardest (using up the most oxygen) is the one doing the task. This is called a blood-oxygenated-level-dependent (BOLD) measure.

fMRI is known for its good spatial resolution: it can figure out precise locations (within one or two millimeters). But it's not so great at temporal resolution: it takes several seconds to get an image, because blood takes a few seconds to flow to brain structures. EEG (see below) has the opposite (good temporal but poor spatial resolution). Some mental events are just too fast for fMRI to capture.

One issue with this is that the whole brain is active all the time. So what they have to do is use a "subtraction task," which is supposed to be unrelated. They subtract the activity of one from the other and look at the differences.

Brain Mapping
There are also efforts to map the neurons of brains. A representation of all of the neurons and their connections is known as a connectome. Connectomes have been completed for two very small species: C. Elegans, a worm with only 302 neurons, and the marine organism Ciona intestinalis.

But the human brain is incredibly complex, with around 100 billion neurons, which is roughly the number of stars in the Milky Way. These neurons have 1,000,000,000,000,000 (that's 10^15) connections. If a dozen or so of 2019's microscopes worked around the clock, it would take thousands of years to map one brain.

Part of the challenge is that tracing neuron links has to be done by hand, looking at images. Computer vision algorithms are getting better at automating this, but as of 2019 they still need to be checked by hand.

The value of a connectome is debated. Some think that it's too detailed, and that more abstract organizational principles are what are needed. Others think it's not detailed enough, and that we need to generate a synaptome that has everything the connectome has, but also information about the nature of the synapses, such as which neurotransmitter receptors there are.

Electroencephalography (EEG)
In EEG, electrodes that measure electrical activity are attached to the scalp. Firing neurons release electrically-charged particles, causing a current. Just like current that comes out of your power outlet, brain currents have electrical cycles.

EEG measures this oscillatory activity in the brain. This rhythmic activity is broken up into "bands." These appear to be important for the brain coordinating and communicating with itself. Low frequency communication is good for synchronization over longer distances (in the brain), and high frequencies are more precise and are used locally.

During wakeful relaxation, you see a lot of alpha waves. When you're concentrating, or problem solving, you get more beta waves, and delta waves are seen during sleep or meditation.

In a typical EEG experiment, some stimulus is presented, and then evoked potentials (EPs, or event-related potentials, or ERPs) are measured. They are either positive or negative (depending on the wave polarity). There is a lot of noise, though, from the environment and other, irrelevant brain processes. Because the noise is random, it can be removed by making measurements over multiple trials and averaging the result. The noise cancels itself out.

P300
The P300 component is something that tends to happen after a surprising stimulus is presented. Scientists can look for P300s to see if the participant’s mind didn’t expect something. It tends to happen about 300 milliseconds after stimulus presentation, but the length of this delay is interpreted as the speed at which the mind is classifying the stimulus. The amplitude of the P300 is interpreted as an estimation of the attentional resources allocated when memory updating is happening.