An Introduction to Cognitive Neuroscience
The chapter Brain and Cognition (Sejnowski & Churchland, 1990) from the book, Foundations of Cognitive Science, is the ideal introduction to the field of cognitive neuroscience. The main aim of this chapter is to elucidate how cognitive neuroscientists study problems of the brain and mind at different computational and structural levels. It also expands on the techniques used to study the brain. Finally, the paper asserts that the goal of cognitive neuroscience is to integrate computational and psychological theories with the neurobiology of the brain.
There are two levels of investigation of problems in cognitive neuroscience, levels of analysis, and levels of organization:
Levels of analysis: This level is drawn from computer science, where the researcher seeks to solve a problem using three steps. These three levels of analysis help researchers understand how to decompose a problem and create a solution.
- Abstract problem analysis: The researcher decomposes the problem into its main parts.
- Algorithmic: The researcher specifies a formal procedure to solve the problem. This may be a formula or a computer algorithm.
- Implementational: The researcher seeks to implement the solution using a physical system.
Levels of organization: In the context of neuroscience, this level details how the nervous system is physically organized. The nervous system is organized into systems, maps, layers and columns, networks, neurons, synapses, and molecules. At each structural level, one can create computational models. Here’s a summary of each structural level:
- Systems: These are widely distributed areas in the brain that connect and measure one sensory modality or functional characteristic such as vision. A system includes many neurons that have minimal effects on each other.
- Maps: Neurons within most systems are arranged in topographic maps. For example, adjacent neurons in the visual field measure adjacent areas of space in the visual world.
- Layers and columns: Brain areas have multiple layers and columns, each responsible for a different function. For example, layer 4 receives sensory input from the thalamus.
- Local networks: These refer to the interconnected web of neurons and synapses in every cubic millimeter of brain tissue. They are measured using single-unit recordings.
- Neurons: A neuron is the most basic unit of processing in the nervous system. Neurons interact with each other through chemical and electrical synapses.
- Synapses: These are gaps between adjacent neurons that help them communicate with each other.
- Molecules: Molecules maintain the integrity of neurons, synapses, and thus the whole brain. In a sense, the entire brain and body are made up of molecules.
The next part of the chapter address how researchers measure these different brain structures. It details the following anatomical and physiological techniques:
- Lesions: The brain can be damaged in various ways, such as through strokes, accidents, tumors, and wounds. In such cases, the researcher can examine the brain using imaging techniques or autopsies and correlate them with behavioral deficits to hypothesize the function of the damaged region. Creating lesions in animals is a prevalent technique to study brain function. However, the lesion technique has limitations, such as different sizes, locations, and premorbid conditions between patients. Researchers can also administer reversible lesions and microlesions using injections of neurotransmitters and other chemicals to study brain function.
- Imaging techniques: This is perhaps the most common way to study brain function at the moment. Structural imaging techniques include computed tomography (CT) and magnetic resonance imaging (MRI) scans. Functional imaging techniques include positron emission tomography (PET) and, most recently, functional magnetic resonance imaging (fMRI), although the latter was not yet used when the chapter was published. Imaging techniques preserve spatial resolution but lack precise temporal resolution.
- Gross electrical and magnetic recordings: These techniques measure signals directly from the scalp. They include electroencephalography (EEG) and magnetoencephalography (MEG). They preserve temporal resolution but lack spatial resolution.
- Single-unit recordings: This involves inserting microelectrodes into the brain to record extracellular potentials. Single-unit recordings preserve spatial and temporal resolution but are infinitely cumbersome to conduct for every neuron in the brain.
Even though there are various methods to study the brain, as highlighted above, there are still many neurobiological constraints when creating cognitive models and hypotheses. It is easy to develop multiple computational models to explain neural processes since many possible solutions exist to a particular problem. The challenge of cognitive neuroscience is to fit these computational models at the algorithmic level to the neurobiology of the brain at the implementational level. Cognitive science aims to explain higher levels of organization, such as systems, while neuroscience addresses the molecular levels of organization. Thus, as cognitive neuroscientists, we must strive to create theories that respect neurobiological constraints.
References:
Sejnowski, T. J. and Churchland, P. S., 1990, Brain and Cognition, In: M. I. Posner (Ed.), Foundations of Cognitive Science, Cambridge, MA: MIT Press
