Learning and Predictive Control in the Ageing Brain

Neuroscientists and psychologists have found that skilled performance depends partly on two forms of mental processing. 'Controlled' processing is dependent upon a part of the brain called the prefrontal cortex which supports higher functions such as decision-making and holding information in mind. This is credited with being flexible, and capable of solving new and unfamiliar problems. However, it is slow, depends on feedback as the relevant task is ongoing, and depends on attention and so fails if overloaded with information. In contrast, there is another system that overcomes some of these limits. 'Automatic' processing depends on a brain structure called the cerebellum, which is known for it's role in performance with minimal awareness (such as riding a bicycle). This is fast, relatively independent of attention, but is inflexible, dependent on people learning to execute tasks ahead of performance, and automatic behaviours are difficult to stop even if important events happen that suggest a different course of action is needed. In older people, the transition from automatic to controlled processing becomes slower with age, and there is a greater reliance on controlled rather than automatic processes which makes older people prone to particular kinds of performance failures that are less frequent in younger people. For example, while driving, drivers aged 60 to 80 years are especially prone to making the wrong decisions at junctions (e.g. roundabouts and crossroads) which makes them far more susceptible than younger drivers aged 18 to 28 years to collisions. We aim to gain a better understanding of how learning aids the transfer from controlled to automatic processing in older and younger age people. Our work will be completed in two parts. In the first, we will use a brain scanning method (functional MRI) to scan people as they perform two different forms of learning. One of these involves the adaptation of a a simple reflex using a well-understood form of learning called classical conditioning, and the other involves a more complex form of trial-and-error learning in which people learn eye movement rules. Both involve learning to use sensory information to instruct a movement and cause activity in different parts of the cerebellum. We will test for evidence that the cerebellum can learn and hold both types of memory in the same way, and expect to see that cerebellar changes will be less pronounced in older than younger subjects. The second part of our project will be focussed on aging and the impact of learning on driver behaviour. We will use a state-of-the-art driving simulator in which older and younger subjects will 'virtually' drive through a series of routes. We will vary the number of times that they visit these roads and see instructions to turn left or right. We will measure behaviour by recording steering behaviour, and we will also use a device to record eye movements (an eye tracker) and will test eye movement performance, alongside steering behaviour, in response to demands of driving. Data will be collected from older and younger people. We will test the hypothesis that eye movement and steering errors will decline more slowly with increasing experience in older compared with younger subjects.

The overarching aim of this work is to gain an understanding of learning-related transitions from controlled to automatic modes of information processing in older and younger age groups. In one workpackage, each subject will be scanned using functional magnetic resonance imaging (fMRI) as they undergo the two forms of learning. In classical eyeblink conditioning, a simple form of cerebellar-dependent motor learning, predictive conditioned stimuli (CSs) that evoke learned responses reliably cause consecutive trial-to-trial phasic activity decreases in cerebellar cortical lobule HVI that are thought to drive the execution of learned responses. We will test for this during human eyeblink conditioning and in the same subjects, we will test whether this same signature of cerebellar plasticity is evoked by CSs in a form of instrumental, trial-and-error oculomotor rule-learning in which becomes increasingly automatic. We expect to observe that such cerebellar changes will be less pronounced in older than younger subjects. The second workpackage will be focussed on aging and the impact of learning on driver behaviour. Our studies will conducted using state-of-the-art driving simulator facilities in which older and younger subjects will 'virtually' navigate a series of routes. We will vary the level of experience gained in each of these routes, and will test oculomotor performance, alongside steering behaviour, in response to predictive stimuli, particularly those that instruct oculomotor and steering decisions associated with approaching junctions. Data will be collected from older and younger subjects in the same age range as that in work package 1. We will test the hypothesis that oculomotor and steering errors will decline more slowly with increasing experience in older compared with younger subjects. We will also measure the extent to which controlled and automatic processing fail to coordinate by studying immunity to distractors associated with increasing automaticity.

Driving is an important ability that helps older people to retain their independence. The number of older drivers aged 60 and above continues to increase in the UK. Although people in this age group are considered to be relatively safe drivers compared with younger age groups, specific decision-making problems render them particularly prone to crash risk on the roads. For example, risk is particularly high at junctions, where the demands related to visual attention and "right of way" decision-making are also high. Older drivers are likely to benefit from evidence-based training that is tailored specifically to their needs, and such training may decrease risks to older drivers.

Here, we aim to understand the mechanisms through which decision-making in older people (particularly in relation to eye movement skills) becomes automatic through learning. We will use brain imaging to understand the neural mechanisms that support the coordination of controlled and automatic eye movement decision-making in older people. We will also use state-of-the-art driving simulation facilities to study the learning of eye movement skills in older people during simulated driving, how eye movement training benefits steering, and the costs and benefits of training to the processing of distracting stimuli.

Our main stakeholders are older drivers, the Department of Transport and two leading charities that represent the interests of older drivers (the Institute of Advanced Motoring, IAM; the Royal Society for the Prevention of Accidents, RoSPA). They will provide input into scientific planning during key stages of our project. We will work with our stakeholders to help them to use our data to tailor assessment and training programmes for older drivers. In the longer term we will also raise awareness among Parliamentarians and will work them to inform policy changes through routes such as the Parliamentary Advisory Committee on Transport Safety and the House of Commons Transport Committee.

Our project will also contribute to capacity building through the training of a postdoctoral research assistant and a PhD student. They will gain valuable experience of impact delivery as part of their scientific training because they will be actively engaged in the delivery of each element of our impact plan.