The tide is changing in our understanding of old age. For a long time, behavioural scientists have thought that old age is associated with cognitive decline such as memory problems, and difficulties in learning and concentration.
Healthy ageing, Ramscar explains, may be nothing more than gaining experience, and then dealing with the consequences of having learnt from that experience:
Older adults’ changing performance reﬂects memory search demands, which escalate as experience grows.
In other words, as people get older, they gather more experiences, they learn more names for things, and they potentially better understand how the social and economic systems around them work – and this makes them slower.
So while youth has the benefit of speed and flexibility, age has the benefit of wisdom and guile … and slowness.
Some of this we already know, even if we’ve never really thought about it in this context. Years of research have shown that older people have larger vocabularies than younger people, other things being equal.
In their paper, Ramscar and associates show that even this we’ve probably underestimated, because older people tend to know a lot of very low frequency words such as “zaftig” and “arroyo” and “byzantine”, words that are difficult to test because there are so many of them. Younger people tend to know fewer of these words.
You can get a sense of this yourself. Have a third party pick up a dictionary and read random “rare” words to you and someone who is either older or younger than you, then see who knows more definitions for these rare words. The research suggests that the older person will know meanings of those rare words when the younger person simply has no idea.
Relating the unrelated
We also know that older people tend to do better in many decision making tasks but Ramscar and colleagues go much further. They show that knowing more has consequences in terms of speed and demonstrate this via a series of analyses involving cognitive models of learning and simple analyses of text.
Using a well-test cognitive model, they show that by simply teaching it more, they can make it slower at recognising certain things (like words). This slowness is characteristic of what many studies find in older individuals.
In another study, Ramscar and colleagues show that the learning impairments may also be due to older people knowing more.
A standard task for this is the paired-associate task. In the paired-associate task a person is asked to remember a set of word pairs, like UP-DOWN and OBEY-INCH.
If they later see OBEY, they should say INCH, for example. Older people often do more poorly overall than younger people when trying to learn many of these pairs.
However, the real signal in this data appears to be that older people are much better at learning the associated pairs, like UP-DOWN, but poorer at learning less typical pairs like OBEY-INCH.
Ramscar and colleagues show that this is predicted from the statistical structure of a lifetime of experience with text. In other words, the older individuals will over-learn common relations, but also learn that unrelated things are … well … unrelated. It’s harder for them to learn these unrelated things – they have a lifetime of experience telling them otherwise.
The message is fairly intuitive. Computers get slower as we store more information on them. Information gets harder to find in libraries for each additional book stored in that library. Libraries are vast and valuable, but they are rarely fast.
Compare that with a little bookshop. You can get in and out quickly, but you may be less likely to find what you’re looking for.
A version of this article was published on Statistical Life.