Evidence is accumulating that age-related cognitive decline is rooted in three related factors: processing speed slows down (because of myelin degradation); the ability to inhibit distractions becomes impaired; working memory capacity is reduced.

A new study adds to this evidence by looking at one particular aspect of age-related cognitive decline: memory search.

The study put 185 adults aged 29-99 (average age 67) through three cognitive tests: a vocabulary test, digit span (a working memory test), and the animal fluency test, in which you name as many animals as you can in one minute.

Typically, in the animal fluency test, people move through semantic categories such as ‘pets’, ‘big cats’, and so on. The best performers are those who move from category to category with optimal timing — i.e., at the point where the category has been sufficiently exhausted that efforts would be better spent on a new one.

Participants recalled on average 17 animal names, with a range from 5 to 33. While there was a decline with age, it wasn’t particularly marked until the 80s (an average of 18.3 for those in their 30s, 17.5 for those in their 60s, 16.5 for the 70s, 12.8 for the 80s, and 10 for the 90s). Digit span did show a decline, but it was not significant (from 17.5 down to 15.3), while vocabulary (consistent with previous research) showed no decline with age.

But all this is by the by — the nub of the experiment was to discover how individuals were searching their memory. This required a quite complicated analysis, which I will not go into, except to mention two important distinctions. The first is between:

• global context cue: activates each item in the active category according to how strong it is (how frequently it has been recalled in the past);
• local context cue: activates each item in relation to its semantic similarity to the previous item recalled.

A further distinction was made between static and dynamic processes: in dynamic models, it is assumed the user switches between local and global search. This, it is further assumed, is because memory is ‘patchy’ – that is, information is represented in clusters. Within a cluster, we use local cues, but to move from one cluster to another, we use global cues.

The point of all this was to determine whether age-related decline in memory search has to do with:

• Reduced processing speed,
• Persisting too long on categories, or
• Inability to maintain focus on local cues (this would relate it back to the inhibition deficit).

By modeling the exact recall patterns, the researchers ascertained that the recall process is indeed dynamic, although the points of transition are not clearly understood. The number of transitions from one cluster to another was negatively correlated with age; it was also strongly positively correlated with performance (number of items recalled). Digit span, assumed to measure ‘cognitive control’, was also negatively correlated with number of transitions, but, as I said, was not significantly correlated with age.

In other words, it appears that there is a qualitative change with age, that increasing age is correlated with increased switching, and reduced cognitive control is behind this — although it doesn’t explain it all (perhaps because we’re still not able to fully measure cognitive control).

At a practical level, the message is that memory search may become less efficient because, as people age, they tend to change categories too frequently, before they have exhausted their full potential. While this may well be a consequence of reduced cognitive control, it seems likely (to me at least) that making a deliberate effort to fight the tendency to move on too quickly will pay dividends for older adults who want to improve their memory retrieval abilities.

Nor is this restricted to older adults — since age appears to be primarily affecting performance through its effects on cognitive control, it is likely that this applies to those with reduced working memory capacity, of any age.