Functional magnetic resonance imaging (fMRI) has transformed cognitive neuroscience since its introduction in the early 1990s. By measuring changes in blood oxygenation associated with neural activity — the blood-oxygen-level-dependent (BOLD) signal — fMRI produces detailed three-dimensional maps of brain activation with spatial resolution of a few millimeters. For cognitive psychology, fMRI has been revolutionary: it allows researchers to observe which brain regions are engaged during perception, attention, memory, language, reasoning, and emotion, providing a bridge between cognitive theory and neural implementation that was previously impossible without studying brain-damaged patients.
Key Structures
- Whole brain
- Default Mode Network — A network of brain regions active during rest, mind-wandering, and self-referential thought — deactivated during demanding external tasks.
Key Functions
- Maps neural activity via BOLD (blood-oxygen-level-dependent) signal.
- used to localize cognitive and sensory functions.
The BOLD Signal
fMRI does not measure neural activity directly but relies on neurovascular coupling — the relationship between neural activity and local blood flow. When a brain region becomes more active, its neurons consume more oxygen and glucose. The vascular system responds by increasing local blood flow, delivering more oxygenated hemoglobin than the neurons consume. Because oxygenated and deoxygenated hemoglobin have different magnetic properties, this hemodynamic response produces a measurable change in the MRI signal. The BOLD response peaks approximately 5–6 seconds after the neural event that triggered it, which limits fMRI's temporal resolution despite its excellent spatial resolution.
Experimental Design
fMRI studies typically use one of two designs. In block designs, conditions are presented in sustained blocks (e.g., 20 seconds of faces alternating with 20 seconds of houses), and the analysis compares average BOLD signal between blocks. In event-related designs, individual trials are presented in randomized order, and the hemodynamic response to each trial type is estimated separately. Event-related designs offer greater flexibility (allowing post hoc sorting by behavioral performance or subjective experience) but require more statistical power. Both approaches rely on comparing activity between conditions rather than measuring absolute levels of brain activation.
Modern fMRI analysis has moved beyond simply asking "where" in the brain. Functional connectivity analysis examines correlations in BOLD signal between brain regions, revealing large-scale networks (the default mode network, dorsal attention network, salience network) that organize brain function. Multi-voxel pattern analysis (MVPA) uses machine learning to decode which stimulus or mental state a person is experiencing from the distributed pattern of brain activity. These approaches have revealed that cognitive functions are supported not by isolated regions but by dynamic interactions among distributed neural networks.
Limitations and Interpretation
Despite its transformative impact, fMRI has important limitations that cognitive psychologists must understand. The BOLD signal is an indirect measure of neural activity, separated from the underlying computation by a sluggish hemodynamic response (~5 seconds to peak). fMRI cannot determine whether a region is performing excitatory or inhibitory processing. The need for expensive equipment, a controlled environment, and participant immobility constrains ecological validity. Perhaps most importantly, finding that a region "activates" during a task does not mean it is necessary for that task — it may reflect a correlated but non-essential process. Converging evidence from lesion studies, brain stimulation, and other methods remains essential for causal interpretation.
Disorders
- Not a disorder treatment
- used diagnostically in pre-surgical epilepsy mapping and tumor planning