The eight-second attention span

The claim is that the average human attention span has fallen to around eight seconds, and that this is shorter than the attention span of a goldfish. It has been used to open conference talks, justify advertising formats, and support arguments about what technology has done to concentration. It travels well because it is short, alarming and comparative.

Following it backwards takes four steps.

1. 2015 onward

   The claim in circulation

   Reported across news outlets, industry publications and presentations, generally without any citation at all, or with a citation to a corporate research report.

2. 2015

   Microsoft Canada, Consumer Insights: Attention Spans

   The report is real and can be retrieved. It carries the figure. It did not generate the figure, and it does not claim to have done so.

3. Attribution given

   Statistic Brain

   A commercial statistics aggregation website, credited in the report as the origin of the number.

4. Terminus

   No retrievable study

   The listed attributions do not hold the data. Journalists who contacted the aggregator, including Simon Maybin for BBC Radio 4's More or Less, were unable to obtain a methodology or a named researcher. No primary source has been produced.

The goldfish

The comparison half of the claim has an even thinner record. No study of goldfish attention producing a figure of nine seconds has been identified. What research on goldfish cognition does exist points the other way: work on associative learning and memory in fish, including studies of trained avoidance responses, indicates retention over periods of weeks rather than seconds.

The deeper problem

Setting provenance aside, the construct itself does not hold. Attention researchers do not treat attention span as a single quantity that can be measured once and expressed as a number of seconds. Gemma Briggs, a psychologist at the Open University who studies attention in applied settings, has described the concept as task-dependent: sustained attention on a monotonous vigilance task, attention during a conversation, and attention while driving are measured differently and are not interchangeable.

There is a real literature on vigilance decrement, running back to Norman Mackworth's clock-monitoring experiments with radar operators in the 1940s, and on sustained attention in cognitive psychology more broadly. None of it produces a general figure for the species. A number stated without a task is not an attention span measurement, because attention is only measurable relative to something being attended to.

Why it persists

The figure survives because it confirms a widely held impression and because the retraction, when it appears, is far less quotable than the claim. Microsoft has not defended the number as its own research. The aggregator that supplied it is no longer a reliable point of reference. Yet the statistic continues to appear, often citing the 2015 report directly, which is precisely the step in the chain where the number was received rather than produced.

Emberline records this filing with the terminus marked. There is no study to evaluate, and no effect size to argue about.