Recording Neural Activity

Electroencephalography (EEG) is a non-invasive real-time method of recording neural activity. EEG uses an elastic cap lined with conductive electrodes that are placed on a human scalp where conductive gel is applied to contact points to reduce scalp-electrode impedance (target < 5kΩ). These electrodes collect extracellular electrical potentials from dipolar pyramidal neurons along the outer layer of the cortex. The propagating signals from the neurons either dominate or cancel out based on dipole orientation relative to the scalp surface. Neurons along gyral crowns stand perpendicular to the scalp as radial dipoles that project parabolic fields vertically to EEG electrodes, whereas neurons along sulcal walls lie parallel to the scalp as tangential dipoles that project laterally to the scalp electrodes. These tangential dipoles cannot project vertically to the scalp electrodes resulting in zero wave reading by the EEG along the sulci (Postle, 2026).

Oscillatory Rhythms and Event-Related Potentials

EEG relies on a correlational method where sensory or cognitive stimuli trigger recordable neural events. The raw EEG data is then spectrally deconstructed into time-frequency oscillatory neural rhythms: gamma, beta, alpha, theta, and delta bands. Invalid indexing of data occurs when the EEG records non-neural ambient signals like alternating current or biological artifacts such as ocular blinking. These artifacts introduce false voltage spikes unrelated to cognitive processing which can obscure true neural responses resulting in construct-irrelevant variance if not removed during preprocessing. Event-Related Potentials (ERPs) are research relevant time-bound snippets that are isolated from the average of raw EEG data across many experimental trials. These ERP waveforms plots of voltage amplitude in microvolts (µV) across time in milliseconds (ms) (Postle, 2026).

Constraints of EEG

There are several constraints that must be inferred when using EEG. First, as the brain is essentially a spherical conductor, a scalp EEG is unable to locate with 100% mathematical certainty where deep within the brain a recorded ERP signal originated. The recording limit of each EEG electrode is approximately 3cm of cortical surface. Second, signals undergo a smearing effect and attenuation as electric current propagates through multiple biological layers before reaching the EEG electrodes. The current must travel from cortical tissue through the three meninges (pia, arachnoid, and dura mater) including the cerebrospinal fluid (CSF), the skull, and ultimately the scalp tissue. The sum of these layers creates an approximate resistance of 181.44Ω with the skull accounting for the majority of this impedance. Third, researchers must rely on tens of thousands of neighboring neurons firing synchronously. Unsynchronized firing leads to ineffective recording as positive and negative potentials of neighboring dipoles cancel each other out.

Application: Visual Working Memory Capacity

A multi-site registered report by Strzelczyk et al. (2026) used EEG to validate visual working memory (VWM) capacity limits of three to four chunks of information (K ≈ 3 − 4). Researchers in ten labs coordinated scalp-recorded Contralateral Delay Activity (CDA), an ERP difference wave. The authors demonstrated that CDA negative voltage amplitude increased across hundreds of human subjects from neural capacity K = 2 to 4 and plateaued between K = 4 to 6 empirically validating the long-postulated theory of a process-level storage ceiling of VWM.