Final Interview Portfolio: Mind-wandering the future

Cognitive neuroscience is the scientific approach to study brain cognition and the mechanisms that underlie it. In a broader sense, it is the most general level to study neuroscience; it’s the culmination of everything else, and it ties cellular neurobiology together with systems neuroscience together with human behavior in order to understand how we think, how we understand, who we are. It’s easy to see, then, how it’s so interdisciplinary, and it makes sense that the lines between traditional neuroscience, psychology, sociology, linguistics, philosophy, and computer science are being blurred when we consider viewing the fields from a higher, broader point of view.

Within this field of cognitive neuroscience lies a specific, niche field of study: mind-wandering. It’s essentially cognitive neuroscience jargon for day-dreaming and attention states, but the way it’s studied is actually quite remarkable, and far different from the simplicity that it evokes. Mind-wandering is difficult to tangify, but some very general questions that the field seeks to answer are: how does attention shift between external cues and internally directed thoughts? What are the pathways, neural networks, and brain areas that are responsible for these shifts and our attention spans? What can we understand mental time-traveling, and how does this phenomenon relate to mind-wandering? What emotions correlate with mind-wandering? Can we predict when we’re mind-wandering, and relate this back to our everyday lives to understand it better?

Hopefully, I’ve piqued your interest, but of course, I’m no expert in the field, and unfortunately, I’m not able to answer these perplexing questions. But Julia Kam might be.  A PhD graduate in Psychology from the University of Vancouver, she’s currently a postdoctoral cognitive neuroscience researcher at the distinguished Knight Lab at UC Berkeley, which is part of the Helen Wills Neuroscience Institute. At first glance, she’s not very imposing. She doesn’t quite seem to match the description of someone who has the qualifications that she does, and she looks like she could almost blend in with the undergraduate college campus. But with her extensive experience in the field of mind-wandering, cognitive neuroscience and psychology, it’s clear to see why she’s able to offer some valuable insight into the nuances, future, and implications of mind-wandering and neuroscience as a whole. I met up with Kam here at her lab space in Berkeley to discuss her background, ongoing work and research, and where she sees the future of her field heading.

Barker Hall at UC Berkeley, home to the Knight Lab and the majority of the Helen Wills Neuroscience Institute

The Interview

Can you give me a quick breakdown of some of your work and some specific projects and research that you’re doing?

There’s a lot of researchers studying mind-wandering and a lot of different ways to study mind-wandering but the way I’ve been studying it mostly looks at how mind-wandering affects the way we process information in the external world. So when I’m talking to you and my mind begins to wander, how does that affect the way I process what you’re saying and what I’m seeing? What I’ve found across multiple experiments is that when we mind wander, we tend to process everything else in the external world to a lesser extent. It’s almost as if there’s this idea of an attention spotlight, and when that spotlight turns toward your own thoughts, everything else gets dampened, until my attention gets redirected back into the physical environment. So that’s how I’ve been studying mind-wandering. What other people have done is to look at the kinds of things that we think about when we’re mind wandering – so for example, if we think about the past, the future, are we likely to be more happy or sad when we’re mind-wandering? There’s also a lot of research that looks at what kind of brain areas that are activated when we’re mind-wandering.

What do you personally find most interesting about the current state of neuroscience and how have we progressed in the last 20 years?

It really comes down to some of the technologies that we have – we’re now equipped with the tools to analyze some really complex data, and we have access to rare data like electrocorticography and single unit recordings. Combing this access to this type of rare data and the technology that we have- which include analytical and statistical developments in the way that we treat these kinds of data sets- it really allows us to get a much better understanding of what’s going on in the brain. What we are kind of lacking though, in regards to all this data and technology, is good theory. It’s almost as if we’re not able to catch up with the times in terms of the technological advancements made in the past 10-20 years. We have a lot of good data and a lot of complicated tools to look at the data, but we don’t really know what to look at at the moment, so having this strong theoretical foundation can help guide us through these massive data sets.

What else is hidden inside these massive, complex data sets?

I understand that one of the projects you’re undertaking is using machine learning to predict mind-wandering. Can you explain to me a little about how that works, and how realistic is it?

Based on my doctoral work, we can see that when we’re mind-wandering, our response to the stimulus that is in the external world is dampened, so that can be used as a neural signature of mind-wandering. What’s unique and what adds to this univariate measure in terms of machine learning is that we get to look at multivariate aspects of the data. The way that machine learning works is that it’s a data-driven approach in which we feed in data that we’ve recorded, and that’s not necessarily processed to a great extent, and what it does is it can – through its algorithm – detect whether mind-wandering or not mind-wandering can show a different pattern of activity. When we feed in all the data, is there some sort of pattern that we don’t see in just one measure that can help differentiate between different attention states? I think that it’s definitely a workable hypothesis. When we move from just one or two measures to the complexity and richness of EEG data, we should be able to differentiate attention states.

Understanding how neural activity can help us differentiate attention states is a common research area in the field – as show by this figure from Dr. Helen Wang at UCSF – but Kam takes it to another level by incorporating machine learning to essentially predict mind-wandering.

So going off script for a second, I remember reading a paper from Gallant’s lab, they had subjects look at a picture, and using fMRI, they were able to reconstruct what they were seeing based on the researchers’ data, and they actually had a pretty rough outline of what the image was in terms of spatial orientation and rough color schemes.  And I also seem to remember hearing a similar project in this lab regarding music. So, do you think it will be possible in terms of your work, to be able to predict what we’re thinking, or something along those lines, one day?

I think that would definitely be a phenomenal finding, but I don’t think we’re quite there yet. The examples that you gave with decoding a particular image, or in our lab with decoding speech or music, the way it works is that you have a specific song in mind, you get the patient to imagine it – like imagine speech or image tunes – and we are quite able to reconstruct their speech or tune just from the electrodes placed over the auditory cortex. So there is a lot of success with decoding and machine learning particular specific images and specific stimuli, but I think it’s a lot trickier with decoding thoughts that are random, since we don’t quite have a template for every thought a person has. So that’s why I started with very gross attention states, or thinking about the past or future, as opposed to anything really specific. So we’re not quite there yet, but in terms of attention states, I’m a little more hopeful.

Researchers at Yale have been able to reconstruct images based on fMRI neural activity. Researchers in the Knight Lab can “hear” music based on EEG neural activity. It’s possible to predict mental states already – can we eventually predict thoughts?

How might future developments and advancements in this field impact our everyday lives?

Going back to mind-wandering, there are apps that some of my colleagues have created that you can download, and it probes you about 8 times throughout the day that will ask you questions like: what are you doing right now? Are you paying attention to what you’re doing? How happy or sad are you? And this gives us a huge amount of data and gives us the opportunity to look into when we’re paying attention to something, if we’ll be happier or less happy, and does this depend on if we’re interested in what we’re doing or more interested in what we’re thinking about?

How might your research in attention and mind wandering impact industries like healthcare or big data?

I think there are a lot of applied contexts – there are a lot of areas where this research can be applied. I’m not sure if education falls under healthcare, but you can imagine a lot of education research focuses on students paying attention. So the machine learning research I was telling you about with eye gaze, a lot of it was done in a classroom context with students, with eye trackers placed in front of their computer screen. And so, one application of it is if we can give students feedback on if they’re mind-wandering (like just a buzz or a tone) because of the data we’ve collected from their  eye movements, then just getting them to be aware is the first step of bringing them back from mind-wandering, so I think that’s a really good application for this area of research. For big data, the example I told you about with the app about attention state, emotional state, interests and thoughts over the task they’re doing, if enough people use the app, it gives you a good idea of general patterns over how people are feeling and what they’re thinking throughout the day. I’m sure Facebook and Google can come up with ways to make use of that data somehow.


I think that a lot of the insight Kam had to offer closely resembles some of the things that I’ve learned in my classes as a cognitive science major. The field itself is actually really new, and truth be told, we’ve only really scratched the surface. Many of the most important and groundbreaking discoveries we’ve made in this field have come from the last few decades, where technology has revolutionized how we approach the research, which is quite a contrast to traditional biology, chemistry, and physics, which are fields so old that even Newton himself found himself young in comparison. Moreover, cognitive science as a field is itself very interdisciplinary. For the first time, we see the lines being blurred between traditional scientific approaches and social sciences. Kam mentioned that the technology we have today is incredibly powerful, but surprisingly, one of the greatest challenges for the future is: how can we use these technological advancements and apply them to our extremely complex data sets? What kind of remarkable studies and discoveries remain buried beneath these complex data? Prior to the 20th century, we simply didn’t have the means to study cognitive neuroscience from a broader standpoint. But now that we have those means, the new challenge is if we can uncover those secrets that we’ve never been able to before. We don’t know what to look at, and the future of the field may very well rest in our ability to understand where to start.

Regarding some of the applications of the field and how this all ties back into us as a  society, Kam also brought up some really interesting points, especially regarding education. Nowadays, we see that technology really has become an integral part of modern-day education, and rightfully so. In many instances, technological advancement can improve overall education experiences. As we’ve covered in class, DreamBox is an educational program that teaches math using artificial intelligence algorithms to personalize math lessons by collecting tens of thousands of data points per student per hour. Kam mentioned that in a similar regard, one of the potential applications of mind-wandering is being able to use machine-learning – in a matter very similar to her own research – to help children detect if and when they are mind-wandering based on neural and alpha wave activity. Intelligent programs like these can not only tailor educational lessons, but also help keep students on task and on track. Just getting them to be aware that they are mind-wandering can help shift their focus and redirect their attention back to external environmental cues, and keep them learning and engaged with the lesson.

Mind-wandering as a whole – and cognitive neuroscience in general – has an enormous potential to help us understand why we behave the way we do, and how we can relate this back to who we are and what we understand. Especially considering how novel the field is, it’s remarkable to think about what the future has to offer.


For reference, the full audio recording can be found below:


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s