Why learning is harder as we get older
Children learn. It’s what they do. And they build themselves over the years from wide-eyed baby to a person that walks and talks and can maybe fix your computer, so it’s no wonder that we have this idea that learning comes so much more easily to them than it does to us. But is it true?
There are two particular areas where children are said to excel: learning language, and learning skills.
Years ago I reported on a 2003 study that challenged the widespread view that young children learn language more easily than anyone older, in regard to vocabulary. Now a new study suggests that the idea doesn’t apply to grammar-learning either.
In the study, 24 Israeli students aged 8, 12, or 21, were given ten daily lessons in a made-up language. A rule in the language — not made explicit to the students — was that verbs were spelled and pronounced differently depending on whether they referred to an animate or inanimate object. In the lessons, the students were asked to listen to a list of correct noun-verb pairs, and then say the correct verb when given further nouns. Two months later, the students were tested on what they remembered.
The young adults were significantly faster at learning and more accurate than the other groups. Moreover, the 8-year-olds never succeeded in transferring the rule to new examples (even when they were given additional training, with the rule made more obvious), while most 12-year-olds and adults scored over 90%, with the adults doing best. It’s also noteworthy (given popular belief) that children's pronunciation was inferior to that of older subjects.
The findings point to the importance of explicit learning, as well as indicating that language skills are not reduced post-puberty, as has been suggested. So why does it seem more difficult for most adults to learn a new language? The problem may lie with interference from the native (or indeed any other) language.
I’ll get back to that. Let’s move on to the related question of procedural memory, or skill learning.
Here’s a study in which we learn something truly fascinating about interference. In the study, 74 young people (aged 9, 12, and 17) were trained on a finger-tapping task, then tested on the two following days. Some of the participants were further tested six weeks later. In a second experiment, 54 similarly-aged people had the same training, but also given an additional training session two hours later, during which the motor sequence to be learned was the reverse of that practiced in the initial session. They were then tested, 24 hours later, on the first sequence.
In the first experiment, all age-groups improved steadily during training, in both speed and accuracy, and showed jumps in performance when tested 24 hours later (such jumps are typical in procedural learning and are referred to as ‘off-line gains’; they are assumed to reflect memory consolidation).
These jumps were maintained or improved at 48 hours, and six weeks. The gains were the same for each age-group, but there was a clear difference between the groups in terms of their starting point, with the older ones performing noticeably better initially. Because the effect of practice was the same for all, the performance difference between each age-group was the same at each point in time.
It is worth emphasizing that performance six weeks after the experience was the same, and sometimes better, despite the lack of practice over that time.
So these results challenge the view that children have an advantage over adults in terms of learning skills, and also demonstrate that children improve “off-line” as adults do, indicating that they too have an effective consolidation phase in motor memory.
But the second experiment is the really interesting bit. You would expect, if you learned one sequence and then learned the reverse, that this would interfere badly with your memory for the first sequence. And so it did, for the 17-year-olds. But not for the 9- and 12-year-olds, who both showed a performance gain at 24 hours, as seen in the first experiment.
Moreover, the better the 17-year-olds became at the reverse sequence, the worse their performance on the initial sequence at the 24-hour test (as you’d expect) — but for the 12-year-olds, the better they were on the reverse sequence, the better they did on the first sequence at the 24-hour test.
What does this mean? Why didn’t interference occur in the pre-pubertal children?
It appears that the consolidation occurring in children is different in some way from that occurring in adults.
There are several possibilities. It may be that the consolidation process becomes, post-puberty, more selective. In the situation where there are several different experiences, priority is given to the more recent. It may also be that consolidation simply occurs faster in children.
One mechanism of change may occur through sleep. The structure of sleep changes during puberty, and we don’t yet know whether consolidation occurs during sleep in children as it does in adults. Another is competition for neural resources (transcription and protein synthesis related factors) during consolidation. It has been suggested that this “competitive maintenance” only fully matures at puberty.
On the other hand, it may have to do with the effects of experience. Interference only occurs when tasks overlap at some point. If children are representing the movement sequences in a more specific, less abstract, way than adults, the sequences may be less likely to use the same neurons (e.g. adults are learning a rule; children are learning two different ways of moving particular fingers). Accordingly, training on the reverse sequence provides additional training in the art of moving these fingers in this way, but doesn’t interfere because the pattern is not the same.
Interference is the bug-bear of learning. Interference may be the key to why learning gets harder the older we get — despite a number of advantages. So let’s explore this a little more.
Here’s a small study in which 14 young adults (average age 20) and 12 older adults (average age 58; range 55-70) learned a motor sequence task requiring them to press the appropriate button when they saw a blue dot appear in one of four positions on the screen. The training included several learnable sequences interspersed with random trials. Participants, however, were not informed of this. There were three blocks of trials during the first session (separated by a 1-2 minute rest), and a fourth block on the second session, 24 hours later.
As expected, younger adults were notably faster in their responses than the older group. Less expected was the fact that the older group showed markedly greater improvement on the learnable sequences than the younger group. However, on the second session, while the younger adults showed the expected off-line gain in performance, indicative of consolidation, the older adults performed at the same level as they had early in the first session.
It should be noted that the average reaction time of the older group in the very last session matched the reaction time of the younger group in the first sessions, demonstrating that, while we may slow down with age, we can counter that with training. The fact that the older adults were noticeably better at learning the sequences may reflect the increases in activation seen in motor regions in normal aging, possibly compensating for decreased activation and atrophy in the hippocampus.
But what’s interesting in this context is this lack of off-line gain.
The same thing was seen in another study comparing younger and older adults, which found that, while the older adults showed improvement in general skill on an implicit sequence-learning task after 12 hours, this improvement had disappeared at 24 hours. Nor was it seen at one week.
So why aren’t these memories being consolidated in the older adults?
(This is not to say that all benefit of the earlier training was lost — the improvement over the second session indicates that some memory was retained. So it may be — and is consistent with what we know about the effects of training in older adults — that more, and perhaps longer, training sessions are needed before older adults can properly consolidate new learning.)
Is this because we become slower to consolidate with age? This harks back to the idea that children suffer less interference because they can consolidate memories more swiftly.
Or perhaps it has to do with the greater interference attendant on the brains of older adults being more richly-connected. A computer model mimicked a decline in language learning as a function of the growth in connectivity in the neural network. This computational model suggests that once connectivity in the parts of the brain responsible for procedural memory slows, learning suffers increasingly from first-language interference.
It may be, of course, that both processes are going on. Greater interference, and slower consolidation.
It may also be that the adult brain becomes more selective in the making of long-term skill memory.
It may also be that these (and other) changes in the adult brain lead to more interaction between information-sets that are further apart (see my recent news item on preventing interference). Thus, if you learn something at ten in the morning, and something else at twelve, your brain can, and will, try to relate the two (which can be good or bad). A child’s brain can’t stretch to encompass that. They would need to be explicitly reminded of the first lesson.
I suspect that all these factors are important, and point to ways in which we should approach learning/teaching differently for pre-pubertal children, young adults, and older adults.
In the case of older adults, it is clear that we need to provide the optimal conditions for consolidation.
I have talked repeatedly about the value of spaced training, distributed training, interleaved training. So it’s interesting to note that studies have found that consolidation of motor memories occurs differently depending on whether training occurs in blocks (each sequence mastered before learning another one) or on a random schedule involving all sequences.
Off-line learning is better when motor skills are learned under a random practice schedule. While blocked practice produces better immediate learning, random practice produces better delayed learning. It appears that a random schedule generates activity across a broad network involving premotor, parietal, sensorimotor and subcortical regions, while learning under the blocked schedule is limited to a more confined area (specifically one particular part of the motor cortex).
This suggests that interleaved practice is even more important for older adults. Although it slows down initial learning (which, remember, was better for older adults compared to younger, so there’s leeway there!), spreading the load across a broader neural network is especially important for those who have some atrophy or impairment in specific regions (as often occurs with age).
Judicious resting during learning may also be of greater benefit for older adults. Consolidation occurs most famously during sleep (and let’s not forget how sleep changes in old age), and also occurs to a lesser level while awake, within a few hours of training. But there is also evidence that a boost in skill learning can occur after rests that only last a few minutes (or even seconds). This phenomenon is distinguished from consolidation (it’s called ‘reminiscence’), because the gains in performance don’t usually endure. However, while in some circumstances it may simply reflect recovery from mental or physical fatigue, in others it may have a more lasting effect.
Evidence for this has come from learning in music. A particularly interesting study involved non-musicians learning a five-key sequence on a digital piano. It found that even 5-minute rests during learning could be beneficial, but only if they occurred at the right time.
In the study, the participants repeated the sequence as fast and accurately as they could during twelve 30-second blocks interspersed with 30-s pauses. A third of the participants had a 5 minute rest between the third and fourth block, while another third had the rest between the ninth and tenth block, and the remaining third had no rest at all. Everyone was re-tested the next day, around 12 hours after training.
Participants showed large improvements during training after either 5-minute rest. However it was only those who were given a rest early in the training that continued to show improvement throughout the training. That is, even though the late-rest group matched their performance on block 10, after this ‘jump’ their performance fell on blocks 11 and 12, while the performance of the early-rest group continued to climb after their jump (at block 4). This group also showed the greatest off-line gain. That is, their performance ‘jumped’ more than that of the other two groups when tested on the following day.
In other words, consolidation was affected by the timing of the rest.
Among the late-rest and no-rest groups, improvement during blocks 4-9 was not as rapid as it had been during the first three blocks. This is a typical pattern during motor learning. It may be, then, that resting early allows processes triggered by repetition to develop fully, rather than becoming attenuated through too much repetition. Thus resting early in practice may allow the faster rate of learning to continue for longer. This in turn results in greater repetition before practice ends, leading to a more stabilized (short-term consolidated) memory, and thus greater overnight (long-term) consolidation.
On the other hand, the short-lasting gain achieved by the late-rest group didn’t affect later learning, but did predict the extent to which performance improved after sleep.
Other improvements to learning may come from reducing interference, and taking cognizance of greater selectivity. In the realm of language learning, for example, it’s argued that successful long-term learning in adults is more and more dependent on explicit learning, declarative knowledge, and its automatization. It may be that, for adults learning a second language, greater importance should be placed on explicit comparison with the native language.
It also seems likely that immersion in the new language is more important for adult learners. The problem is that every time you return to your native language, you’re encouraging interference (something to which, as we have seen, children may be far less susceptible).
In sum, as we get older, interference becomes more of an issue. To counter this, we need to be more thoughtful about planning our learning.
For more about the recently reported research into the difference between children's and adults' language learning, see
Brown, R. M., & Robertson E. M. (2007). Off-Line Processing: Reciprocal Interactions between Declarative and Procedural Memories. The Journal of Neuroscience. 27(39), 10468 - 10475.
Brown, R. M., Robertson E. M., & Press D. Z. (2009). Sequence Skill Acquisition and Off-Line Learning in Normal Aging. PLoS ONE. 4(8), e6683 - e6683.
Cash, C. D. (2009). Effects of Early and Late Rest Intervals on Performance and Overnight Consolidation of a Keyboard Sequence. Journal of Research in Music Education. 57(3), 252 - 266.
DeKeyser, R., Monner, D., Hwang, S-O, Morini, G. & Vatz, K. 2011. Qualitative differences in second language memory as a function of late learning. Presented at the International Congress for the Study of Child Language in Montreal, Canada.
Dorfberger, S., Adi-Japha E., & Karni A. (2007). Reduced Susceptibility to Interference in the Consolidation of Motor Memory before Adolescence. PLoS ONE. 2(2), e240 - e240.
Ferman, S., & Karni A. (2010). No Childhood Advantage in the Acquisition of Skill in Using an Artificial Language Rule. PLoS ONE. 5(10), e13648 - e13648.
Ferman, S. & Karni, A. 2011. Adults outperform children in acquiring a language skill: Evidence from learning an artificial morphological rule in different conditions. Presented at the International Congress for the Study of Child Language in Montreal, Canada.
Karni, A. 2011. A critical look at ‘critical periods’ in skill acquisition: from motor sequences to language skills. Presented at the International Congress for the Study of Child Language in Montreal, Canada.
Nemeth, D., & Janacsek K. (2010). The Dynamics of Implicit Skill Consolidation in Young and Elderly Adults. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences. 66B, 15 - 22.
Robertson, E. M., Press D. Z., & Pascual-Leone A. (2005). Off-Line Learning and the Primary Motor Cortex. The Journal of Neuroscience. 25(27), 6372 - 6378.
Stambaugh, L. A. (2011). When Repetition Isn’t the Best Practice Strategy: Effects of Blocked and Random Practice Schedules. Journal of Research in Music Education. 58(4), 368 - 383.
Steele, C. J., & Penhune V. B. (2010). Specific Increases within Global Decreases: A Functional Magnetic Resonance Imaging Investigation of Five Days of Motor Sequence Learning. The Journal of Neuroscience. 30(24), 8332 - 8341.
Wymbs, N. F., & Grafton S. T. (2009). Neural Substrates of Practice Structure That Support Future Off-Line Learning. Journal of Neurophysiology. 102(4), 2462 - 2476.