Tag Archives: cognitive science

The Cognitive Science Age

namib

Complex patterns in the Namib desert resemble neural networks.

The history of science and technology is often delineated by paradigm shifts. A paradigm shift is a fundamental change in how we view the world and our relationship with it. The big paradigm shifts are sometimes even referred to as an “age” or a “revolution”. The Space Age is a perfect example. The middle of the 20th Century saw not only an incredible increase in public awareness of space and space travel, but many of the industrial and technical advances that we now take for granted were byproducts of the Space Age. 

The Cognitive Science Age

It’s probably cliche to write this but I believe we are at the beginning of a new age, and a new, profound paradigm shift. I think we’re well into the Cognitive Science Age. I’m not sure anyone calls it that, but I think that is what truly defines the current era. And I also think that an understanding of Cognitive Science is essential for understanding our relationships with the world and with each other. 

I say this because in the 21st century, artificial intelligence, machine learning, and deep learning are now being fully realized. Every day, computers are solving problems, making decisions, and making accurate predictions about the future…about our future. Algorithms decide our behaviours in more ways that we realize. We look forward to autonomous vehicles that will depend of the simultaneous operation of many computers and algorithms. Machines will (and have) become central to almost everything.

And this is a product of Cognitive Science. As cognitive scientists, this new age is our idea, our modern Prometheus.

Cognitive Science 

Cognitive Science is an interdisciplinary field that first emerged in the 1950s and 1960s and sought to study cognition, or information processing, as its own area of study rather than as a strictly human psychological concept. As a new field, it drew from Cognitive Psychology, Philosphy, Linguistics, Economics, Computer Science, Neuroscience, and Anthropology. Although people still tend to work and train in those more established traditional fields, it seems to me that society as a whole is in debt to the interdisciplinary nature of Cognitive Science. And although it is a very diverse field, the most important aspect in my view is the connection between biology, computation, and behaviour.

The Influence of Biology

A dominant force in modern life is the algorithm, as computational engine to process information and make predictions. Learning algorithms take in information, learn to make associations, make predictions from those associations, and then adapt and change. This is referred to as machine learning, but the key here is that machines learn biologically,

For example, the algorithm (Hebbian Learning) that drives machine learning was discovered by the psychologist and neuroscientist Donald Hebb at McGill university. Hebb’s book on the The Organization of Behaviour  in 1949 is one of the most important books written in this field and explained how neurons learn associations. This concept was refined mathematically by the Cognitive Scientists Marvin Minsky, David Rumlehart, James McLelland, Geoff Hinton, and many others. The advances we see now in machine learning and deep learning are a result of Cognitive Scientists learning how to adapt and build computer algorithms to match algorithms already seen in neurobiology. This is a critical point: It’s not just that computers can learn, but that the learning and adaptability of these systems is grounded in an understanding of neuroscience. That’s the advantage of an interdisciplinary approach.

The Influence of Behaviour 

As another example, the theoretical grounding for the AI revolution was developed by Allen Newell (a computer scientist) and Herbert Simon (an economist). Their work in the 1950s-1970 to understand human decision making and problem solving and how to model it mathematically is provided a computational approach that was grounded in an understanding of human behaviour. Again, this an advantage of the interdisciplinary approach afforded by Cognitive Science. 

The Influence of Algorithms on our Society 

Perhaps one of the most salient and immediately present ways to see the influence of Cognitive Science is in the algorithms that drive the many products that we use online. Google is many things, but at its heart, it is a search algorithm and a way to organize the knowledge in the world so that the information that a user needs can be found. The basic ideas of knowledge representation that underlie Google’s categorization of knowledge were explored early on by Cognitive Scientists like Eleanor Rosch and John Anderson in the 1970s and 1980s. 

Or consider Facebook. The company runs and designs a sophisticated algorithm that learns about what you value and makes suggestions about what you want to see more of. Or, maybe more accurately, it makes suggestions for what the algorithm predicts will help you to expand your Facebook network… predictions for what will make you use Facebook more. 

In both of these cases, Google and Facebook, the algorithms are learning to connect the information that they acquire from the user, from you, with the existing knowledge in the system to make predictions that are useful and adaptive for the users, so that the users will provide more information to the system, so that it can refine its algorithm and acquire more information, and so on. As the network grows, it seeks to become more adaptive, more effective, and more knowledgeable. This is what your brain does, too. It causes you to engage in behaviour that seeks information to refine its ability to predict and adapts. 

These networks and algorithms are societal minds; They serve the same role for society that our own network of neurons serves our body. Indeed, these algorithms can even  change society. This is something that some people fear. 

Are Fears of the Future Well Founded?

When tech CEOs and politicians worry about the dangers of AI, I think that idea is at the core of their worry. The idea that the algorithms to which we entrust increasingly more of our decision making are altering our behaviour to serve the algorithm in the same way that our brain alters our behaviour to serve our own minds and body is somethings that strikes many as unsettling and unstoppable. I think these fears are founded and unavoidable, but like any new age or paradigm shift, we should continue to approach and understand this from scientific and humanist directions. 

The Legacy of Cognitive Science

The breakthroughs of the 20th and 21st centuries arose as a result of exploring learning algorithms in biology, the instantiation of those algorithms in increasingly more powerful computers, and the relationship of both of these concepts to behaviour. The technological improvements in computing and neuroscience have enabled these ideas to become a dominant force in the modern world. Fear of a future dominated by non-human algorithms and intelligence may be unavoidable at times but and understanding of Cognitive Science is crucial to being able to survive and adapt.

 

The fluidity of thought

Knowing something about the basic functional architecture of the brain is helpful in understanding the organization of the mind and in understanding how we think and behave. But when we talk about the brain, it’s nearly impossible to do so without using conceptual metaphors (when we talk about most things, it’s impossible to do so without metaphors). 

Conceptual metaphor theory is a broad theory of language and thinking from the extraordinary linguist George Lakoff. One of the basic ideas is that we think about things and organize the world into concepts in ways that correspond to how we talk about them. It’s not just that language directs thought (that’s Whorf’s idea), but that these two things are linked and our language also provides a window into how we think about things. 

Probably the most common metaphor for the brain is the “brain is a computer” metaphor, but there are other, older ideas.

The hydraulic brain

One interesting metaphor for brain and mind is the hydraulic metaphor. This probably goes back at least to Descartes (and probably earlier), who advocated a model of neural function whereby basic functions were governed by a series of tubes carrying “spirits” or vital fluids. In Descartes model, higher order thinking was handled by a separate mind that was not quite in the body. You might laugh at the ideas of brain tubes, but this idea seems quite reasonable as a theory from an era when bodily fluids were the most obvious indicators of health, sickness, and simply being alive: blood, discharge, urine, pus, bile, and other fluids are all indicators of things either working well or not working well. And when they stop, you stop. In Descartes time, these were the primary ways to understand the human body. So in the absence of other information about how thoughts and cognition occur it makes sense that early philosophers and physiologists would make an initial guess that thoughts in the brain are also a function of fluids.

Metaphors for thinking

This idea, no longer endorsed, lives on in our language in the conceptual metaphors we use to talk about the brain and mind. We often talk about cognition and thinking as information “flowing” as in the same way that fluid might flow. We have common expressions in English like the “stream of consciousness” or “waves of anxiety”, “deep thinking”, “shallow thinking”, ideas that “come to the surface”, and memories that come “flooding back” when you encounter an old friend. These all have their roots (“roots” is another conceptual metaphor of a different kind!) in the older idea that thinking and brain function are controlled by the flow of fluids through the tubes in the brain.

In the modern era, it sis still common to discuss neural activation as a “flow of information”. We might say that information “flows downstream”, or that there is a “cascade” of neural activity. Of course we don’t really mean that neural activation and cognition are flowing like water, but like so many metaphors it’s just impossible to describe things without using these expressions and in doing so, activating the common, conceptual metaphor that thinking is a fluid process.

There are other metaphors as well (like the electricity metaphor, behaviours being “hard wired”, getting “wires crossed”, an idea that “lights up”) but I think the hydraulic metaphor is my favourite because it captures the idea that cognition is fluid. We can dip our toes in the stream or hold back floods. And as you can seen from earlier posts, I have something of a soft spot for river metaphors.

 

 

A Curated Reading List

Fact: I do not read enough of the literature any more. I don’t really read anything. I read manuscripts that I am reviewing, but that’s not really sufficient to stay abreast of the field. I assign readings for classes, to grad students, and trainees and we may discuss current trends. This is great for lab, but for me the effect is something like me saying to my lab “read this and tell me what happened”. And I read twitter.

But I always have a list of things I want to read. What better way to work through these papers than to blog about them, right?

So this the first instalment of “Paul’s Curated Reading List”. I’m going to focus on cognitive science approaches to categorization and classification behaviour. That is my primary field, and the one I most want to stay abreast of. In each instalment, I’ll pick a paper that was published in the last few months, a preprint, or a classic. I’ll read it, summarize it, and critique. I’m not looking to go after anyone or promote anyone. I just want to stay up to date. I’ll have a new instalment on a regular basis (once every other week, once a month, etc.). I’m doing this for me.

So without further introduction, here is Reading List Item #1…

Smith, J. D., Jamani, S., Boomer, J., & Church, B. A. (2018). One-back reinforcement dissociates implicit-procedural and explicit-declarative category learning. Memory & Cognition,46(2), 261–273.

Background

This paper was published on line last fall but was just published officially in Feb of 2018. I came across it this morning and I was looking at the “Table of Contents” email from Memory & Cognition. Full disclosure, the first author was my grad advisor from 1995-2000, though we have’t collaborated since then (save for a chapter). He’s now at Georgia State and has done a lot of fascinating work on metacognition in non-human primates.

The article describes a single study on classification/category learning. The authors are working within a multiple systems approach of category learning. According to this framework, a verbally-mediated, explicit system learns categories by trying to abstract use a rule, and a procedurally-mediated, implicit system learns categories by Stimulus Response (S-R) association. Both systems have well-specified neural underpinnings. These two systems work together but sometimes they are in competition. I know this theory well and have published quite of few papers on the topic. So of course, I wanted to read this one.

A common paradigm in this field is to introduce a manipulation that is predicted to impair or enhance one of the systems and leave the other unharmed in order to create a behavioural dissociation. The interference in this paper was a 1-back feedback manipulation. In one condition, participants received feedback right after their decision and in another, they received feedback about their decision on the previous trial. Smith et al. reasoned that the feedback delay would disrupt the S-R learning mechanism of the procedural/implicit system, because it would interfere with the temporal congruity stimulus and response. It should have less of an effect on the explicit system, since learners can use working memory to verbalize the rule they used and the response they made.

Methods

In the experiment, Smith et al. taught people to classify a large set (480) of visual stimuli that varied along two perceptual dimensions into two categories. You get 480 trials, and on each trial you see shape, make a decision, get feedback, and see another shape, and so on. The stimuli themselves are rectangles that vary in terms of size (dimension 1) and pixel density (dimension two). The figure below shows examples of range. There was no fixed set of exemplars, but “each participant received his or her own sample of randomly selected category exemplars appropriate to the assigned task”.

1F890860-F5FB-4D01-81DD-CFB2D765A31B

They used a 2 by 2 design with two between-subject factors. The first factor was category set. Participants learned either a rule based category (RB) in which a single dimension (size or density) creates an easily-verbalized rule, or an information integration category (II) in which both dimensions need to be integrated at a pre decisional stage. This II category can’t be learned very easily by a verbal rule and many studies have suggested it’s being learned by the procedural system. The figure below shows how many hundreds of individual exemplars would be divided into two categories for each of the each of the category sets (RB and II).

3B0F0A79-A9D8-4CA4-A68E-4D350908BA90

The second factor was feedback. After each decision, you either received feedback right after you made a decisions (0Back) or you received feedback one trial later (1Back). This creates a heavier task demand, so it should make it harder to learn the RB categories at first because the task creates a heavier working memory load. But it should interfere with II learning by the procedural system because the 1-Back disturbs the S-R association.

Results

So what did they find? The learning data are plotted below and suggest that the 1Back feedback made it harder to learn the RB categories at first, and seemed to hurt the II categories at the end. The 3-way ANOVA (Category X Feedback X Block) provided evidence to that effect, but it’s not an overwhelming effect. Smith et al.’s decision to focus a follow up analysis on the final block was not very convincing. Essentially, they compared means and 95% CIs for the final block for each of the four cells and found that performance in the two RB conditions did not differ, but performance in the two II conditions did. Does that mean that the feedback was disrupting the feedback? I’m not sure. Maybe participants in that condition (II-1Back) were just getting weary of a very demanding task. A visual inspection of the data seems to support that alternative conclusion as well.  Exploring the linear trends might have been a stronger approach.

1C1FA11F-E898-435A-845B-B890030342CB

The second analysis was a bit more convincing. They fit each subject’s data with a rule model and an II model. Each model tried to account for each subject’s final 100 trials. This is pretty easy to do and you are just looking to see which model provides the most likely account of the data. You can then plot the best fitting model. For subjects who learned the RB category, the optimal rule should be the vertical partition and for the II category, the optimal model is the diagonal partition.

As seen the figure below, the feedback did not change the strategy very much for subjects who learned the RB categories. Panel (a) and (b) show that the best-fitting model was usually a rule based one (the vertical partition). The story is different for subjects learning II categories. First, there is way more variation in best fitting model. Second, very few subjects in the 1-back condition (d) show evidence of the using the optimal rule (the diagonal partition).

39925827-64C8-40ED-923F-2309265CD7FB

Conclusions

Smith et al concluded: “We predicted that 1-Back reinforcement would disable associative, reinforcement-driven learning and the II category-learning processes that depend on it. This disabling seems to have been complete” But that’s a strong conclusion. Too strong. Based on the modelling, the more measured conclusion seems to be that about 7-8 of the 30 subjects in the II-0Back condition learned the optimal rule (the diagonal) compared to about 1 subject in the II-1Back. Maybe a just handful of keener’s ended up in the II-0Back and learned the complex structure? It’s not easy to say. There is some evidence in favour of Smith et al’s conclusion but its not at all clear.

I still enjoyed reading the paper. The task design is clever, and the predictions flow logically from the theory (which is very important).It’s incremental work. It adds to the literature on the multiple systems theory but does not (in my opinion) rule out a single-system approach. But I wish they had done a second study as an internal replication to explore the stability of the result. Or maybe a second study with the same category structure but different stimuli. It’s incremental work. It adds to the literature on the multiple systems theory but does not (in my opinion) rule out a single-system approach.

Tune in a few weeks for the next instalment. Follow my blog if you like and check the tag to see the full list. As the list grows, I may create a better structure for these, too.