Why My Cognitive Science Degree Was A Great Foundation For Data Science and Machine Learning

Coming into the machine learning and data science field from a cognitive science background was daunting – but ultimately provided a…

Personal Tales Into Data Science

I segued into the domain of data science and machine learning harboring a lot of uncertainty and insecurity. I had endless curiosity and excitement – doe-eyed and optimistic. But ringing in the back of my mind was the insecurity that I didn’t come from any of the traditional backgrounds, for example, computer science, statistics, or business. Instead, I graduated with a bachelor’s in cognitive science. However, as time passed and my experience grew, an idea began to slowly unravel – perhaps, my background provided a much more solid foundation than I had initially anticipated.

First Things First – What Is Cognitive Science?

Whenever someone asks about my major, I tend to recite:

"Cognitive Science is an interdisciplinary field of neuroscience, artificial intelligence, computer science, philosophy, psychology, linguistics, and anthropology."

This doesn’t explain much and is often met with a slow nod and an uncertain _"…okay."

Consequently, I’d like to take a moment to express what I’ve studied and why I’ve developed gratitude for the framework it provided me.

The Mystique of the Mind

I began my studies with a strong disposition to learn about the mind; one of the strongest avenues of discovery is through cases in which cognition fails.

I’ve learned that some people cannot recognize faces. Others see sounds visually through color. Some actually feel pain when they see people in pain. Others have an intense desire to amputate their limbs because they believe it’s an intruding entity. And some have been woken up with their own hand uncontrollably choking them. Most importantly, I’ve become aware that there is a biological impairment that is associated with each disorder. And these biological impairments have treatments.

Before we knew anything about the existence of these disorders, we would categorize anyone that deviated from normalcy as "insane". For example, those with prosopagnosia may have trouble recognizing their children or coworkers. This means if they were to walk past someone they know but cannot recognize, and fail to greet them, they may appear to be rude.

Or some individuals with body integrity identity disorder have a desire to amputate a major limb because they feel as though it is impeding. Consequently, onlookers would deem them "crazy" instead of considering the underlying physical impairments that have contributed to it.

So What Exactly is Cognitive Science and Why Is It Interdisciplinary?

However, mental dysfunction was just one aspect of my studies. My major, in short, revolves around "understanding cognition". In other words, this mysterious ability we have to think, reason, and acquire knowledge.

In this discipline, I was provided with a new breadth of self-awareness and avenues of contemplation as to what the human experience entails. Prior to the field of cognitive science, we studied this topic using strictly introspection and psychology, OR strictly neuroscience and biology.

But cognitive science proposes that in order to understand the complexity of the mind, we cannot rely on a single discipline to have all of the answers. We need to understand both the internal states of the mind as well as the anatomy of the brain – and this is still not enough. We need to understand society, culture, language, modern technology, and everything else that may contribute to developing our ability to think.

To be more precise, my studies enabled me to delve into consciousness, dreams, memory, attention, existence, the nervous system, society, culture, the impact of drugs on the brain, the process of developing computer systems that marginally mimics the brain’s neural networks, as well as the other aspects of evolving technology.

So Where Does Data Science and Machine Learning Come In?

I’ve listed below a few ways I’ve seen the world of cognitive science and the world of data science/machine learning collide.

Neural Networks and Neuroscience: Neural networks are inspired by the human brain’s neural circuitry. It was insightful to study neurons, connectivity, and how information gets propagated forward before learning about the mathematical representation of this biological phenomenon. Although the field of deep learning introduces novel jargon, the underlying biological processes that these mathematical models were inspired by are a lot more complicated; the network architectures are simplified mathematical representations. Moreover, a lot of variations in deep learning architectures are also inspired by a component of the biological counterpart or an idea to mimic human cognition. It has been a lot easier to conceptualize these models and concepts after having a foundation of the more complicated, biological inspirations.
Human-Computer Interaction and Data Visualization/Storytelling: A critical part of understanding the mind is understanding how humans capture and retain information. In cognitive science, we study the underlying processes of memory and attention. We examine how attention is captured and what humans are likely to focus on, and how we naturally group visual cues to understand an overarching story. Furthermore, we investigate the differences in cognition between cultures and how lifestyle and tradition affect the way we think. This background enables me to think critically about visualizations. Who is the audience and what do these colors represent for this group? Does my visualization direct the attention of the audience to what I am trying to convey? Is there too much clutter that will distract my audience from quickly and intuitively understanding the message? How much information can a human understand at once?
Statistical Analysis: With any scientific endeavor, comes statistical analysis. Science deals with the examination of data, probability assessments, and prediction analysis. From the statistics courses I’ve completed, I’ve carried with me a healthy skepticism for every new piece of information or statistic I come across. I enjoy asking questions along the lines of: " Who were the subjects in the study? How was sampling done? Did the experimental design align with the initial question? Does the experimental design enable us to make the broader conclusions with confidence? What can we do with this information and is it ethical?" This same skepticism is applied to machine learning problems. It’s important to take a methodological and comprehensive approach, especially if the deployment of our models holds societal ramifications.
Computational Social Science and Network Analysis: There is a framework within cognitive science that explores distributed cognition which, according to Wikipedia, "refers to a process in which cognitive resources are shared socially in order to extend individual cognitive resources". The goal is to analyze how cognition arises outside of the confines of an individual, which includes exploring interactions between individuals. My computational social science class translated quite directly to network analysis. My class investigated social interactions and behavioral interactions through simulation and modeling. In order to analyze the degrees of connectivity, we applied network analysis techniques. These same techniques are ubiquitous within the realm of data science.
Neural Signal Processing and General Time Series Analysis: The goal of neural signal processing is to decompose neural signals in order to extract information about how the brain represents and transmits information. Since these neural signals have a temporal component, we can represent them as time-series data. Consequently, I was able to apply the same techniques used in that class to other time-series data sets.

While I’ve elaborated on only 5 connections between my cognitive science and machine learning studies, the list doesn’t end there. A few other shared topics include calculus, linear algebra, A/B testing, data ethics, etc. It was heartwarming to slowly unravel the realization that my former studies didn’t place me at a disadvantage; rather, it provided a solid framework to build on.

Imposter Syndrome

I began my transition to data science and machine learning with an overwhelming feeling of imposter syndrome. Despite my uncertainty, I never felt regret for studying the mind. The study of the mind developed within me a desire to serve the community of those suffering from neurological impairments or battling with mental health issues. Meanwhile, the study of data science and machine learning birthed an excitement to utilize the computational power we have to serve that community.

Lastly, it’s worth noting that while imposter syndrome can be paralyzing, in many instances it had fueled me to learn faster and intentionally. I began with fear that these topics would fly over my head, only to find that my background surprisingly allowed me to make an intuitive connection. Now that I have a "zoomed out" lens, the fear and uncertainty from not coming from a "traditional background" is replaced with pure gratitude.

Advice

If anyone reading this is feeling that same uncertainty coming from a non-traditional background, know that you are not alone. Many data scientists and machine learning practitioners do not come from traditional backgrounds. Remember that the study involves tools that we can apply to any domain and your path will enable you to make unique connections that can help you solidify new concepts.

Lastly, don’t let the feeling of imposter syndrome overwhelm you. Rather, allow it to fuel you with the motivation and intention necessary to keep up to date with the ever-changing landscape of data science and machine learning.