sort circuit

Featured as part of the Dear World Project exhibition, a collaboration with cognitive scientist Alex Hopkins and artist Alban Low and myself. Exhibited at Stour Space on Fish Island in Hackney and curated by Julia Vogl. All photos by Jake Fairnie

 ‘The history of Dear World Project…’

In 2018-19 we visited festivals and events around London and East Anglia asking members of the public to write a postcard about what was currently on their mind. Over 800 postcards were collected, and the public were invited to read other postcards and try and categorise the writers content with one label. Were they talking about their ‘mood’? Were they talking about how much ‘effort’ they felt they were making? How ‘social’ they felt they should be? Categorising thoughts and feelings is hard. This is what we try and do every day when we try to communicate our internal experience to others. But this can be difficult when one description can mean many things. One person who says, “I haven’t been feeling myself recently” could be experiencing lack of concentration and energy. Someone else may be describing feeling very anxious. In the clinic two people may receive the same diagnosis of depression – but one may have very low mood and insomnia while another may have a lack of interest in activities and be sleeping a lot.

‘…to the future of mental health research.’

The postcards contained rich and varied statements on how people were thinking and feeling. We wondered, could we try and group these together to understand more about people’s state of mind? Are there any similarities in how people are feeling? We fed the stamps from the postcards to a machine learning algorithm that tries to group together in ‘clusters’ postcards that are similar to each other. The algorithm built us 3 clusters, and told us how much each postcard belonged to each one. We translated these clusters into a visual code where background, chip and circle all represent different groups of postcards. These create new insights into how our life experience can be categorised, by understanding the commonalities between ourselves. Researchers are using techniques like these with big data to try and uncover new patterns in symptoms. But do we risk losing the individual using this approach? What do we gain, and what do we lose by using categories and labels?