Broadbent's early selection theory

Meaning: Broadbent proposed that sensory input from many channels enters a brief sensory store in parallel, but a filter lets one channel through to a limited-capacity stage where meaning is analyzed. The filter selects on physical characteristics such as pitch, loudness, or spatial location. Unselected channels are blocked before semantic analysis, so little of their content is processed or remembered.

On the MCAT: The usual setup is a dichotic listening or shadowing task. Participants wear headphones, repeat the message in one ear, and are later asked about the other ear. They can report physical changes in the ignored ear, such as a switch in voice or tone, but little of its content. That result supports early selection. A result in which people notice their own name in the ignored ear (the cocktail party effect) challenges it. Two later models answer that challenge. Treisman's attenuation model says unattended input is weakened, not fully blocked, so important words can still break through. Deutsch and Deutsch's late selection model says all input is analyzed for meaning and selection comes afterward.

Common confusion: "Early" refers to where selection happens in the processing sequence, namely before meaning is analyzed, and not to a time in the experiment. Broadbent's filter blocks the unselected channel, while Treisman's attenuator turns it down.