Periodic Reporting for period 1 - PSYNAT (The Psychological Benefits of Interacting with Nature)
Período documentado: 2021-04-01 hasta 2023-03-31
Over the past two years, I have worked on the proposed project and I do believe we are steps closer to understanding how and why nature is beneficial to our cognitive abilties, though this work will need to be continued for years to come in order to truly understand the nuances of the relationship between nature exposure and well-being.
To test the effects of isolation from nature, I downloaded the images from the Aesthetic Visual Analysis (AVA) database, which includes thousands of images spanning a wide range of categories. To put a neural network "in lockdown", I implemented an existing neural network architecture using PyTorch and trained the network on only images labelled as "indoor". I also trained versions only on "landscape" (i.e. nature/outdoor) scenes, only on "urban" scenes, and a baseline version including all three scene types. We found that, compared to baseline, the "urban" and "indoor" networks resulted in lower aesthetic predictions on test images. In other words, these networks seemingly became "depressed" and did not enjoy nature images as much as the baseline network. The opposite was the case for the nature-only network. Overall, these results suggest a negative long-term effect of isolation from nature - namely, people would lose their ability to enjoy typically aesthetically pleasing environments, leading to an increase of depressive symptoms. I presented this work at the European Conference on Visual Perception (ECVP) in Nijmegen Netherlands in August 2022. I am currently preparing the manuscript for submission. However, due to the quickly-evolving nature of the field of deep learning/AI, there are now better aesthetics neural networks and image databases available that I would like to use before finalizing the manuscript.
However, the existing research had not clearly defined "soft fascination" or how/why nature captures attention “moderately”. Thus, we turned to theories of aesthetic pleasure. The Infovore Hypothesis posits that people enjoy seeking out information, and thus derive pleasure from scenes in which there is a great amount of novel and interpretable information to decipher. Similarly, within a predictive coding framework, the brain is constantly generating predictions about its environment, therefore we seek to gather new information to reduce prediction errors about the world. The Pleasure-Interest Model of Aesthetic Liking further theorizes that if there are no prediction errors (i.e. disfluency) to reduce, the resulting experience is boredom. The advantage of these aesthetic theories over the ART is that they explain how a scene can be described as simple but not boring, or interesting but not overwhelming (i.e. the “moderate” cases).
By researching the cognitive benefits of nature through the lens of aesthetics research, I was able to find a novel link between attentional focus and improvements on a working memory task. Here, I was able to show that "soft fascination" may represent a decoupling between attention and aesthetics. While we typically pay more attention to the things we like, I found that in order for an environment to be beneficial, it should be pleasant without capturing more attention.