On the Symbiosis of Remembering, Forgetting and Learning

While I immensely enjoyed the theme of the theme of Bjork's research, I found myself somewhat skeptical about some of the assumptions made by the author. Mainly, I found the assertion that we (society, in general) have an imperative to have accurate models of how we learn, to be somewhat unfounded. Having some experience with social psychology, I am aware that knowing how cognitive biases, methods of persuasion, normative influences, etc. work, doesn't do much to alter the effects of these mechanisms. The examples given, like the inaccurate ability of people to predict their ability to pair words correctly, might be somewhat to blame for me skepticism. The tests used in the studies mentioned in Bjork's paper aren't particularly compelling or relatable.

Having taken computer science classes, where much of the description of how a computer's memory works was related to us as almost a one-to-one with our own brain's memory mechanisms (inaccurately it would seem), I was interested to read Bjork's analysis of the reverse understanding could negatively impact people's perception of their own memory and learning capacities. I wonder if the younger population, one that was raised using computational technologies, might be particularly susceptible to thinking metaphorically and incorrectly about their learning, remembering, and forgetting processes.

Lastly, some of the implications of Bjork's paper are stress inducing, particularly the understanding that the more one memory is retrieved, the stronger it will be and the weaker competing memories could be. While reading this, I considered how eye-witness testimony pointing to one assailant could continuously get strengthened by testimony, while counter-testimonial information might increasingly get weakened. However, my reading of the text wasn't all sour and dour, there was an element of the text regarding the reverse fundamental attribution error, if I may, that I thought was really interesting and should be potentially revelatory. The passage spoke to how we tend to over-emphasize the role of the person in learning and memory retrieval, as opposed to the situational factors that play a considerable role. I think that many people, for a variety of reasons, are type-cast as smart or strong learners (or the opposite) based solely on their ability to retrieve information, when their homelife or situational stress could be playing a role in their ability to perform these functions.

I really enjoyed this paper on remembering, forgetting, and learning. Following the discussion points about recreating context for a user when retrieval is necessary at a later time, I think this is the broader sense of affordances. The knowledge of how to open a door is not stored in our head, but instead, stored in the affordance of the door. When designing context aware systems, we can figure out the information people are trying to retrieve at certain context, and either store that information to direct present to the user when needed, or use context cues to help the user retrieve that information.

Another point I find interesting is abstraction and forgetting. This is one part that human is still much better than machine learning systems. We can learn the high-level abstraction not with a million data points, but with very few exemplars. By forgetting the specific example of solution, we got to learn more important problem solving skills. Some of these are used in deep learning systems, like LSTM uses forget gate and surprisingly performs much better. I believe to achieve real AI, more and more of human cognition theories will need to be applied into the design of AI system, even from the lowest level of computation units.

+1 to mmadaio's comments about conflicting purposes of testing in the US education system.

On a somewhat sadder note, the retrieval framework helps us understand why trauma can be so terrible for people; the tendency to replay an embarrassing, sad, or traumatic memory over and over prevents it from fading away. This is similar to what Joselyn wrote about in the courtroom example, but instead of creating shared (potentially false) memories, we're talking about continued impact of negative ones on a person.

I'm also somewhat interested in generational differences in perception of our own mental processes (such as memory), as Joselyn talked about. Are we more likely to view ourselves more mechanistically now than young adults fifty years ago? What are the social implications of that?

I'm currently reading a book called "Make it Stick" by Peter Brown. It pretty much looks into the all the published literature of how people learn. One of the concepts that stood out the most to me is, if you want people to truly learn something, contextualize it by attaching it to something they already know in the past. For example, if you are trying to teach someone about gravity, tell them that the thing that pulls people to the ground when you jump down from a mango tree (don't judge, that's how my Nigerian teacher taught me) is gravity. That way, you build a system of connecting things to each other, which greatly enhances retrieval.

With regards to the paper, I'm with Bjork on being against the idea of learning as a positive only outcome. I'm just not entirely convinced that it is necessary for people to design mechanisms that prevent them from forgetting. There are some things that it is just not worth my time storing/remembering, and others where it is critical. I cannot imagine a time where computers will not be available to assist humans with retrieval of things like phone numbers and addresses (which in my opinion, it is better at). I'm also not entirely convinced that forgetting enhances abstraction, maybe I do these things subconsciously, but I would assume that it makes abstraction worse?

I also find the concept of "desirable difficulties" to be very interesting. However, to add to Michael M's point on exams, I feel the high-stakes and high-stress nature of exam does not necessarily have a negative impact on the students' motivation, as they also get extra sense of achievement on surviving/acing the exams.

To add to Francesca's point, I don't think the authors mean that we'll have to forget 3D geometry to learn how to program. But I do hope the authors can give a more clear description and definition on when will forgetting actually promote learning and in what situation will preventing forgetting nullify learning.

Some of Bjork's claims are indeed a bit controversial. At times, it almost seems it could go either way. He mentioned a study in which students studying twice in different contexts performed better than those both in the same context. But isn't it possible that some learning might have been associated to the context and returning to the same context could promote recall? The triangular relationship between learning, remembering and forgetting, in my opinion, is probably not governed by a fixed equation. My hypothesis is such relationship has a fair amount of uncertainty. For example, forgetting can enable learning, and can also hinder learning, as it is also dependent on the individuals, and affected by other factors.