PSY 402 Topic 4 DQ 1: Early vs Late Selection

The single confusion that wrecks this post is thinking early and late describe how quickly you respond. They describe where in the processing stream unattended information is discarded. Early selection, in Broadbent's filter model, says selection happens before meaning is extracted: a filter operates on physical properties such as which ear the sound arrived at or what pitch it had, and everything else is blocked before it is identified. Late selection, associated with Deutsch and Deutsch, says all input is processed to the level of meaning and selection happens afterwards, at the point of response or awareness. Treisman's attenuation model sits between them and is worth mentioning: the filter turns unattended channels down rather than off, so a sufficiently important word can still get through. That middle position is where the evidence has largely settled. Get the attribution right as you go, since the attenuation model is frequently credited to the wrong author in summaries.

Now the examples, which the prompt asks for explicitly and which most posts under-serve. For early selection, use dichotic listening: people shadowing one ear notice a change from speech to a tone in the other ear, but cannot report what language was spoken there — physical properties survive, meaning does not. For late selection, use the cocktail party effect: your own name spoken across a noisy room captures attention, which is impossible unless the unattended stream was processed far enough for the name to be recognised. State what each example proves and what it does not, because neither settles the argument on its own. The productive conclusion is that selection is flexible: with high perceptual load it looks early, with low load it looks late, which is why a single fixed filter position was always going to be the wrong answer. Load is measurable: raise the attended task's demands and the same participant shifts from late to early selection.