What waits for a question

A smell brings back a kitchen you have not thought about in thirty years. A quiz question pulls up the capital of a country you could not have named a minute earlier. In both cases the knowledge was there all along, and nothing in your day had asked for it. Which raises an odd question: how much of what you know is like that, stored and never called?

It is harder to answer than it looks, because the only way to find a piece of knowledge is to trigger it. Every count is a test, and every test brings cues of its own. Memory research has found three side doors into the unprompted part, and each shows it is large.

The first is the cue. In 1966 Endel Tulving and Zena Pearlstone had people learn lists of up to 48 words drawn from named categories, then tested some by free recall and others with the category names as prompts.¹ Cued recall beat free recall, especially for the longer lists, and when the free-recall group was given the category names afterwards, they recalled more, mostly from whole categories they had missed entirely.¹ Nothing about storage differed between the groups; only the question did.¹ "Any particular test of memory," as the psychologist John Kihlstrom summarises their view, "measures only that information which is accessible under the conditions prevailing at the time of retrieval."¹

The second is recognition, which offers the item itself as the cue and finds far more. In a 2008 study, people who had looked through pictures of 2,500 objects over five and a half hours could pick out the one they had seen 87% of the time, even when the alternative was the same object in a different state or pose.² Nobody could recall 2,500 objects unprompted. Vocabulary shows the same gap in ordinary life. An average 20-year-old American knows about 42,000 lemmas in the sense of recognising them as words, and the ability to use words oneself is estimated at less than half of that.³ Most of a vocabulary waits for someone else to say the word first.

The third door finds what the other two miss. In a 1978 study, people learned word-number pairs; four weeks later some pairs could be neither recalled nor recognised, yet these "forgotten" pairs were relearned faster than new ones.⁴ Something was still stored that no ordinary test could reach.

Thomas Landauer tried to put a number on the whole. From the rates at which people take in and keep information, measured with recall and recognition tests on reading, pictures, words and music, he estimated the content of an adult memory at around a billion bits; one of his estimates of the net gain over 70 years is 1.4 billion.⁵ Even that is a count of what tests can reach.

So no one can say what share of a life's knowledge is never triggered, and the reason is built into the question: whatever we count, we have triggered by counting it. What the side doors agree on is that the unprompted part is the small part. How much a person seems to know, to others and to themselves, then depends partly on the questions their life happens to ask. Two people with the same memory, living among different cues, would know different amounts by any test anyone could give them.