Building rich theory piece by piece

The winners of this year’s British Psychological Society Doctoral Research Award and Spearman Medal have been delving in to some of the essential processes of human experience.

Dr Aidan Horner (University of York: pictured left) is the winner of the 2018 Spearman Medal, which is awarded by the Society each year to someone who has produced outstanding work within eight years of completing a PhD. Horner’s research has revealed the psychological and neural mechanisms behind our ability to look back in time and re-experience previous life events.

When Horner joined Neil Burgess’s lab at UCL as a postdoctoral researcher he wanted to explore some of the basic computational mechanisms that underlie episodic memory using a mixture of brain imaging, statistical modelling and behavioural testing. Specifically he wanted to find out more about all-or-none retrieval – how we can recall a full, complex event from a single cue (such as revisiting the location of a first date), or recall none of that event.

There have been hypotheses based on computational models that attempt to explain what the hippocampus does during this type of retrieval – it is suggested different elements of a memory, for example the subject, object and location, are stored in disparate areas across the neocortex and are combined within the hippocampus. Similarly when we experience one of the ‘cues’ that make up part of a whole memory of a complex event, the hippocampus aids in the retrieval of all the other parts of the memory – the music that played in the background or the person we were with, for example. Horner’s behavioural experiments and fMRI studies confirmed these neural processes when participants recalled previously learned groups of stimuli comprising a location, person and object.

Horner has also investigated grid cells, which sit in the medial temporal lobe and fire in a spatially dependent way – playing an important role in representing space. These are established in the literature to be involved in navigation, both in humans and rodents, but Horner found that grid cells fire not only during actual navigation but during imagined navigation as well.

‘I come from a psychology background, but I’m also interested in neuroscience and what individual neurons are doing in the medial temporal lobe, and trying to link between those things can be extremely hard. It’s very rewarding when you can go from some high-level behaviour and subjective experience all the way down to individual neurons in the brain and how they’re potentially wired up – but it’s certainly challenging trying to do that.’

Over the coming years Horner is hoping to find out more about what happens to a memory once it’s encoded and the ways in which memories degrade over time. He said he currently had two hypotheses as to how complex memories are forgotten: ‘We know if we learn the location, person and object then that’s stored in a relatively coherent representation that allows for this holistic retrieval. The question is what happens as we forget? Do these memories fall out as a whole – do we lose all the information associated with location, person and object at one time? Or do they start to fragment over time? Perhaps we forget the person or the object first. We’ve been trying to tease apart those two potential ways in which we might forget these episodic memories and have been using the statistical modelling approaches I’ve developed to tackle those questions.’

Dr Neil Bramley’s PhD research, completed at University College London, has examined how humans make sense of the causal structure of the world. His work has been awarded the British Psychological Society’s Award for Outstanding Doctoral Research Contributions to Psychology.

Bramley, now a Moore-Sloan postdoctoral associate in the Centre for Data Science at New York University, has introduced new ideas and approaches to the study of causal cognition in both theory and methodology – his work combines careful experimentation with sophisticated mathematical and computational modelling.

Originally a philosopher, Bramley later became fascinated by human consciousness and embodied cognition and set out to explore how people both pick up basic concepts about how the world works and learn about new, unknown concepts. He has drawn on the scientific process itself as an analogy for the ways people intuitively interact with their surroundings to discover more about causality: ‘It turns out people are quite good at systematically manipulating things while controlling for confounding variables that they don’t understand yet, targeting and isolating something they wonder about and doing something that’s like a mini experiment. I try to use information theory to describe what’s going on when they act on the world and are trying to learn something and whether it’s got the properties of a good experiment.’

While many models within psychology attempt to explain how people learn based on lab experiments they often ignore the true complexity of real-world human experience – that we are often faced with hundreds of potential variables when we are attempting to make sense of the world. Bramley’s PhD work tackled this complexity and his models suggest we build up complex beliefs in an incremental way.

‘How do you end up with a big, rich theory about how lots of things relate to each other? I think we build it up piece by piece where we focus on one little sub problem at a time but when we do so we lean on our surrounding beliefs for support. If we gradually solve little sub problems, always assuming that our assumptions are kind of correct, for long enough we can gradually change the whole thing and gradually find a better overall theory.’

More recently, Bramley has become interested in grammars within thinking and planning, or the suggestion that higher-level cognition may have a compositional structure rather like a language. ‘There seems to be something compositional there in the sense that you can construct really interesting ideas from putting together simpler ideas in an unusual way. There’s this new idea that’s big at MIT and NYU which is rethinking learning as a kind of self-programming where what you’re trying to do is program yourself to have a good recipe for solving a problem, or building concepts out of parts in the same way you might write a small bit of code if you were a programmer.’