In a special issue of Neural Networks devoted to attention,it might seem at the very least perverse to write about pre-attentive processing. However, it is only by understanding thepower of pre-attentive processes that we can understand whatattention has to work with, and indeed against.In this note,we consider the critical function of the selectionfrom an entire visual scene of visual locations or objects fordetailed or attentive processing. Such selection of some placesat the expense of others is necessary because attention hasonly a meagre capacity (estimated at just 40 bits/s; Sziklai(1956)). The same bottleneck implies that the selection processitself cannot generally be attentive (bar an explicit cue or someother form of effective guidance). Fortunately, pre-attentivemechanisms,operatinginparallelacrosstheentirevisualinput,seen most prominently in pop-out (with, for instance, a red dotpopping out among the green ones, or a vertical bar amonghorizontal bars), operate to offer a vastly simpler substrate ofsalience for attentional selection (Julesz, 1981; Neisser, 1967;Treisman & Gelade, 1980). Concomitantly, tasks that requirethe selection of non-cued locations that are not pre-attentivelysalient, are much more daunting.What clues are there to the mechanisms employed by pre-attentive selection? The computational requirements indicate(a) fast parallel processing, (b) a spatial substrate for selection,and (c) a complex featural basis of effects such as pop-out. These particularly implicate V1, as the largest, andretinotopic, cortical visual area, with many cells tuned todifferent visual feature(s) responding at each location in visualspace. Physiological data (Allman, Miezin, & McGuinness,1985; Knierim & Van Essen, 1992; Nothdurft, Gallant, & VanEssen, 1999; Sillito, Grieve, Jones, Cudeiro, & Davis, 1995)suggest that V1 contributes to pop-out, since a V1 neuron’s