What is it that makes something meaningful, and why do we assign certain meanings to certain things? Is there inherent meaning in the world, or is it only the meaning we construct for ourselves? Is there a difference? In my thesis project, I sought to investigate these questions, looking at how we perceive the world, and how this perception allows us to construct and create meaning, whether it be real or imagined.
Through my investigation, I began to see how the process of meaning-making is a form of creativity, and the interpretation of randomness can function as a way to generate new ideas; ideas that could not exist without the seed of the random starting point. This focus on creativity led me to develop two creative tools, both of which facilitate creative thought through the exploration of randomness within a structure. The focus with both of these approaches was not to explicitly teach the participants about the process of meaning-making, but rather to guide them into exploring the question themselves.
The Google DeepDream project began as a way for researchers to investigate how neural networks compartmentalize and categorize imagery, but ultimately results in the creation of imagery that is both fascinating and repulsive. How did images like these come to exist?
In this case, the neural networks had been trained on a set of labelled images, allowing them to identify and categorize previously unseen data based on the information learned in training. This gives a neural network a basic sense of the properties that make up different objects. For example, in processing labelled training sets of photos of bananas, a neural network learns the visual cues that signal “banana-ness”. What the Google DeepDream project found was that this process not only allows networks to sort imagery, but it also allows them to generate it.
The experiments that most interested me were the ones that started from complete randomness, to give way to something meaningful. Given an image of random noise, the network is told to look for a banana, gradually adjusting the image and then reprocessing it until “banana-like” features become recognizable. From here, the process is run again on the same image, until something resembling a banana appears. The interesting part about this process is the inherent meaninglessness of the original image of random noise. The network was told to look for something and found it, creating imagery meaningful even to human eyes. But this meaning was generated from nothing but the “desire” to find something; there was nothing inherently “banana-like” about the initial image, but this does not change the way we assign meaning to the outcome.
The experiments that most interested me were the ones that started from images of random noise. Given an image of random noise, the network is told to look for a banana, gradually adjusting the image and then reprocessing it until “banana-like” features become recognizable. From here, the process is run again on the same image, until something resembling a banana appears. The interesting part about this process is the inherent meaninglessness of the original image of random noise. The network was told to look for something and found it, creating imagery meaningful even to human eyes. But this meaning was generated from nothing but the “desire” to find something; there was nothing inherently “banana-like” about the image of random noise.
From this starting point, I investigated the ways in which human thought is similiar to the workings of neural networks. We are constantly interpreting meaning and assigning significance to things that are inherently random: clouds in the sky, the burn marks on a piece of bread, ink blots on a sheet of paper, “faces” on the surface of the moon.
This interpretation of meaning is a result of our perception, and our brains’ processing of reality, and has no bearing on reality itself. We see and understand things that are only real because of our attention; They only exist because we are there to see them. And in this way, what is not real, becomes real.
In my research into the creation of meaning from meaninglessness, I explored the various ways in which this process occurs in nature, and in the mind.
Pareidolia is the perception in the mind of a meaning or pattern that does not exist, such as seeing a face on Mars, or an animal in the clouds. The human brain is wired to look for patterns, with the most common type of perception being that of faces. Beyond facial recognition, there exists other kinds of perception which go beyond the purely evolutionary advantage of being able to recognize faces and process emotions.
Recognized as a culturally universal phenomenon, humans are known to perceive religious imagery in stimuli where there is no intentional representation being made. Those who are religious are more likely to experience this kind of perception, as they have a confirmation bias that primes them to notice these particular kinds of patterns in everyday life. Religious belief also causes people to attribute meaning to these perceived images, such as taking them as a sign confirming the existence of a higher power. In this way, perceptions of religious imagery are a unique kind of Pareidolia that go beyond an isolated attribution of meaning; Rather, their perceived meaning contributes to a larger belief structure. It’s not just seeing a face in a rock and finding it strange, but rather seeing the face and finding it as support for the existence of God.
A constructed vehicle for the encouragement of meaning creation is the Rorscharch test, which values subjective interpretations above reality, and uses these interpretations to gain insight into the mind of the participant. Here, one is encouraged to interpret meaning into the meaningless, and what is perceived is more important than what is objectively “true”. The Rorscharch test creates a structure within which meaning interpretation can occur. This format served as the basis for both of my studio projects.
In learning about pareidolia, it became evident to me that perception is based largely on our expectations and past experiences. We are more likely to see what we expect, and in seeing what we expect, it is our expectations that construct our realities.
The Disintegration of the Persistence of Memory by Salvador Dali, 1952 - 1954
The Great Masturbator by Salvador Dali, 1929
The surrealist painter Salvador Dali recognized the importance of archetypes to the human perception of the world. Archetypes create a sense of order and stability, allowing us to categorize items so as to respond to them with prior knowledge of what they are. In order to know that something is irrational, we first must compare it to our rational archetypes. Dali developed the Paranoiac Critical Method to explore a way of recognizing the irrational without first comparing it to the rational. Within this menthod, iterative association occurs, wherein one object morphs seemingly nonsensically into another, and the only rationalization needed is that Dali’s mind somehow found a connection between the two. This concept of archetypes is connected to the method used by systems to recognize images; As humans use archetypes, neural networks use a labelled training set of images against which they compare data.