A major focus is on trying to build computational models that explain how memory breaks down in older adults and those with specific memory impairments (amnesia). These models try to simulate the functions of the hippocampus, a tiny brain structure known to be centrally involved in learning and memory. His research on decision making studies differentiates between conscious, deliberative choices and unconscious, intuitive ones.

He has authored or edited several books, including Straight Choices: The Psychology of Decision Making (with Ben Newell and David Lagnado) and the Sage 5-volume collection of classic research articles, The Psychology of Learning. He is a Professor of Psychology and head of the Division of Psychology and Language Sciences at UCL.

Our knowledge and memories are highly organised and integrating new knowledge into this existing network not only speeds up learning but also greatly boosts the likelihood that it will be remembered later on.

A simple illustration of this is that a random list of words like diamond, horse, orange, rabbit, purple, emerald… is harder to remember than a list comprising the same words grouped into taxonomic categories diamond, emerald… horse, rabbit… orange, purple… Organizing words into common groups allows us to see connections between them and to use simple cues (‘animals’) to help us recall them later.

The extraordinary performance of memory champions provides a spectacular example of the power of elaborate encoding. When Johannes Mallow set the current world record for memorizing 501 random digits in 5 minutes, he did so by grouping them into 3-digit chunks each of which triggered a specific mental image, which he mentally connected together. He didn’t just try to remember the digits in their basic form, he elaborated them via visualization.

When Memrise helps you to connect a word (such as ‘aburrido’) and its translation (‘boring’) by suggesting a ‘mem’ (it’s boring to eat a burrito with every meal), it’s helping you to create an elaborate encoding.

The first demonstrations of the power of testing go back almost a century but had little impact on educational practice. Indeed as recently as 1989 the testing effect was described as “not gone but nearly forgotten” by one researcher. The person most responsible for current wider appreciation of the benefits of testing is Roddy Roediger from Washington University in St Louis, USA, who in a long series of both laboratory and classroom studies has demonstrated that tests (casa = ?) boost memory far more than simply re-studying the material (casa = house) for the same amount of time. Indeed this research shows that re-studying is often virtually ineffective. A recent major project undertaken by Roediger and his colleagues demonstrated that tests, in the form of brief quizzes, boosted learning of scientific concepts and principles for 8th grade students in a classroom context. As a result of these tests, students scored higher in their school exams. Spending the same amount of time on reviewing the course material rather than being quizzed had no such beneficial effect.

One of the most striking things about the benefit of testing is that people are typically unaware of it. Research has found that people often believe they will benefit more from an opportunity to restudy or review some information than they will from being tested on it, despite the converse usually being the case.

Why is testing so beneficial for memory? Despite the huge amount of research on this question, there is no agreed answer, and indeed there may be several explanatory factors which are not mutually exclusive. One way to think about testing as a learning event is from the perspective of what Robert Bjork calls ‘desirable difficulty’, the principle that factors that make learning harder are often paradoxically beneficial for long-term retention. Recalling an answer in a test is often effortful, and it may be that this effort is necessary and desirable to strengthen what has been learned. Consistent with this idea, evidence suggests that greater benefits accrue from harder tests, for instance from a recall test with no clues as to the answer (casa = ?) compared with a recall test with part of the answer supplied (casa = h___?).

Another possible explanation is that repeated tests enable the key information to become separated from any specific contexts and instead to be associated with a wider range of retrieval cues. Jeffrey Karpicke and his colleagues have recently provided strong support for this idea.

Memrise adds another element to the mix, which is to vary the ways in which your memory is tested. Thus, on one occasion, you may be asked simply to recall an answer (casa = ?) whereas on another occasion you may have to pick the correct answer from 4 alternatives. This variation serves to keep your learning interesting, but probably also boosts the benefits compared to using just a single type of test. Laboratory research documenting this possibility is rather limited, however.

Another reason why tests may boost memory beyond what can be achieved by restudying is that errors can be made on tests. Provided that feedback in the form of the correct answer is immediately provided, the commission of errors is not unduly harmful and may indeed promote effective learning.