New review punctures the myth that now is three seconds long

By Emma Young

When you say it's gonna happen now

When exactly do you mean?

Ask a psychologist the answer to this question – posed in this case by Morrissey in The Smiths song, How soon is now? – and she might reply "within the next three seconds".

The idea that "now", also known as the "subjective present", is constrained within this time limit has proved popular. But a new evaluation in Psychological Bulletin of dozens of research papers on everything from embraces and reading poetry to tapping along to a beat concludes that there's no good evidence for it. Our experience of the present cannot, it seems, be so strictly defined.

Plenty of events that we perceive as happening at the same time actually don't quite co-occur, of course. An intriguing illustration of this comes, for example, from the neurological patient known in the literature as PH. This man hears people's voices before he registers the movement of their lips. At first, he thought the TV wasn't correctly dubbed, but then he realised the same thing was happening with people he was talking to. Evaluation of his case led the researchers involved to conclude that there are various clocks in the brain, and PH has developed problems with the coordination of those linked to vision and hearing.

Still, there's only about a 200-millisecond time difference between our brain's processing of the sight of someone's lips moving and the sound of their voice. So how big can a discrepancy get for a healthy brain not only to cease perceiving events as simultaneous, but to stop binding perceptions together into "the present"?

In his new critique of the three-second theory, Peter White at Cardiff University first considers evidence put forward from experiments into duration perception: if someone sounds a tone for two seconds, say, or eight seconds, how good is somebody else at holding down a button for the same length of time?

Ernst Pöppel at the Ludwig-Maximillian University of Munich, Germany, has claimed that at about three seconds or under, we're accurate, but beyond that, we under-estimate durations – which, he's argued, supports the idea of two different brain processing modes: one for events that last less than three seconds, and one for events that take longer. This supports the idea that "now" lasts up to three seconds, he argues. In fact, a clear-cut boundary between performance below or over three-second durations does not exist, argues White. Rather, it seems that as the duration of a tone gradually gets longer, our performance tends to gradually worsen, too, and there's no clear evidence of a shift in accuracy at the three-second mark.

How about work investigating how well we can tap along to a beat? While some studies support the idea that we're relatively accurate when the beat intervals are less than three seconds (which, again, Pöppel argues, suggests that we automatically process and integrate events that take three seconds or less as chunks of "now"), others just don't, White points out.

Pöppel, and a surprising number of others, have also argued that human behaviours are structured in roughly three-second units. Wiping or rubbing something, pouring liquid from one pot to another, or reading a line of poetry, for example have been claimed to take approximately three seconds. Again, this is taken as evidence that our brains deal in three-second windows of time, with anything that happens within that time frame being viewed as a single event. (An analysis of footage of athletes and coaches embracing at the Beijing Olympics even found that embraces tended to last an average of 3.17 seconds.)

But there's a significant methodological problem with at least some of the research, argues White: "How is a behavioural unit identified?" White cites research finding that when lay volunteers are shown videos or animations of human behaviour – such as an actor using a keyboard or chasing someone – and are asked to break these behaviours down into the smallest units that seem natural, their segments generally last a lot longer than three seconds.

In fact, White argues, while the idea of a three-second subjective present is "simple and eye-catching, which probably contributes to its popularity", it's not accurate. There is, instead, he argues, an "envelope of integration" – a flexible period of time in which the brain incorporates data into an overall representation "that has no fixed time limit, but in any case extends well beyond the three-second mark."