Processing latencies for coherent, high level percepts in vision are at least 100 ms and possibly as much as 500 ms. Processing latencies are less in other modalities, but still significant. This seems to imply that perception lags behind reality by an amount equal to the processing latency. It has been proposed that the brain can compensate for perceptual processing latencies by using the most recent available information to extrapolate forward, thereby constructing a model of what the world beyond the senses is like now. The present paper reviews several lines of evidence relating to this hypothesis, including the flash-lag effect, motion-induced position shifts, representational momentum, static visual illusions, and motion extrapolation at the retina. There are alternative explanations for most of the results but there are some findings for which no competing explanation has yet been proposed. Collectively, the evidence for extrapolation to the present is suggestive but not yet conclusive. An alternative account of compensation for processing latencies, based on the hypothesis of rapid emergence of percepts, is proposed. Keywords: perceived present moment; extrapolation to the present; predictive coding; flashlag effect; perceptual processing latencies. Word count: Main text: 16,921 Extrapolation to the present 3 Is the perceived present a predictive model of the objective present? All of us have the feeling that we live in the present moment; that is, in terms of perception, that we perceive the world outside our brains as it is now. I shall hereafter use the term "perceived present" to refer to the collective set of perceptual information that seems to us to represent the outside world as it is now. In fact, operation of peripheral sensors, neural transduction and perceptual processing of input information take time, so it can be argued that our conscious percepts are out of date by an amount equal to the combined latency of those things (hereafter just "processing latency", for convenience). 1 That is, we are in effect perceiving a moment that is in the past by however long it takes for perceptual processing to generate the percept of that moment. An alternative possibility is that the brain uses the processed information, not to represent the recent past, but to predict the present moment. In that case, conscious percepts would be the contents of the predictive model of the present moment, generated from the most recent available information. The aim of this paper is to review the evidence concerning that possibility. I shall start by briefly describing the processing latency problem, and I shall then proceed to a review of several lines of research that have been taken as evidence for the predictive model hypothesis, which I shall call the extrapolation to the present hypothesis. At the end I shall propose an alternative account, based on the hypothesis that conscious percepts begin to emerge no more than 100-150 ms after stimulus onset (ASO). Several kinds of adjustment to the temporal location of thing...