Mirror Neuron
Description:
Virtually anything in this world survives if enough people regard it highly, or consider it valuable. What if an artwork would be able to increase its own chances of existence by making sure it is being liked? What if it would be able to assume what we like and modify itself accordingly?
Mirror Neuron is a new media artwork currently in development. The artwork is based on the human mechanism of mirroring in social interactions. It follows concepts from psychology, namely simulation theory and theory of mind, applied to computers in an art context. The name hints to the so-called mirror neuron that, it is speculated, might provide the basis for theory of mind in the brain.
The artwork is also a reflection on the current status of technology when it comes to making assumptions and generalizations about people – and the accuracy of it. It also points to the fact that perhaps more and more images we will see digitally will become highly personalized.
Shown here is the first prototype of the piece. It makes use of the YOLOv4 object detection model to get high-level information from a camera pointed at the viewer, then uses this information to drive changes in real-time of color and luminosity in the parametric visual. This means that the artwork changes itself according to what it “sees”.
Further prototypes will include a different computer vision model that can detect gender and age, a database of research data regarding color preference in people of different ages and genders, and a more complex interaction between the data and the parametric visual.
Artwork by:
Aurora Mititelu
Year:
2021
Exhibited at:
A.R.E Generative Art Fest – December 2021, Berlin DE
Future Development Goals:
In developing the project further, I am interested in having a better understanding of all its aspects: technical, theoretical as well as aesthetic.
Regarding the theoretical approach, I want to dive deeper into the mechanisms of human social interaction. These include building rapport, mirroring, as well as judging and inferring other people’s behaviors. Some of these aspects are explained by the theory of mind, which talks about the capacity to understand other people by attributing mental states to them. I want to understand how that might apply to machines and speculate on the idea of machine “social” behavior in human interaction.
Technically, I want to better understand what are the current capabilities of Computer Vision and its different models, as well as when and how are they being used. It is important to observe how accurate these technologies are in their applications and what type of high-level data they can handle. I am also interested in what kind of research data is available on human preference, especially regarding color and visual features of an image, as well as where and how is this information being used: what industries or academic institutions produce this data and for what purposes? In what applications is it being used today?
Lastly, in terms of aesthetics, I want to get a glimpse of what kind of features of an image are considered pleasing for different people, starting with color. Why is an image beautiful, whe do we like an image? If such data would be available, I would be interested to use localized information for each country the installation would be exhibited in, to have a feeling of preference across cultures.
