Blurts

There is an opinion that those of us who teach psychology often have to tolerate. I would translate this opinion as the belief that the strategies we spend our time researching could be identified by common sense. Some assume that what we teach and test is about the theory and research behind such practical applications and these topics are not especially valuable because most folks would function in their daily lives according to common sense and achieve very similar results. My specialization, which includes courses intended to prepare future teachers but also talking to students about their learning strategies, frequently confronts this attitude. Both future teachers and college students have experienced the educational process as learners for many years and so often feel well-informed when it comes to the strategies they should apply for success. Is this the case?

Many examples demonstrate that the behavior of learners is far from optimal. An obvious example would be how students spend their time. Studies of massed versus distributed practice clearly show the long-term benefit of distributing the investment of study time. Yet, cramming for exams is common. I understand that the issue here is often one of motivation rather than the understanding of learning. Simply put, long-term retention may not be what many students prioritize. Their goal may be to maximize their performance on tomorrow’s test with far less interest in whether they can recall the content next year. If students understand this and don’t care, then cramming may not represent a good test of the claim I am making. The problem I intend to describe here seems different, but even cramming ignores the reality that courses build on each other and background knowledge has a significant impact on success with future related courses.

When it comes to studying and even to what has become my most recent interest, nonacademic note-taking and the application of knowledge gleaned from personal reading, there are examples more easily related to the lack of understanding of how learning works. Such examples come from situations in which a learner decides between two strategies that require the same commitment of time and chooses to focus on the less productive strategy. When such examples can be identified, this is not the end of what researchers want to learn. Why are poor choices common? Are learners unaware of their poor choice and the issue might then be resolved through greater awareness or are other more challenging issues involved? To get back to the issue I first identified, why doesn’t common sense solve this challenge? 

The value of retrieval practice represents an example of the situation I have just described and this phenomenon has received a great deal of recent attention. An easy way of operationalizing one task that involves retrieval practice is to describe this task as testing. The purpose is not testing for a grade, but just for benefits generated by the experience of testing. This task might be controlled by an educator who spends some class time having students use clickers or their cell phones to respond to questions based on assigned reading or classroom presentations. Testing might be self-imposed with one common strategy being the use of learner-created “flash cards”. 

So, to implement the type of research scenario I have just claimed demonstrates that learners make poor choices, it is necessary to first know that one strategy is more productive than the other. A simplified description of such research would involve two groups assigned at random to conditions that either involve the review of notes and book highlights or the same content with some of the time spent completing test items. When both groups work for the same amount of time, the group that included retrieval practice (testing) performs better on a delayed test of what was learned. 

Once established, the focus becomes what do learners do when study “in the wild” or when offered the option of a study session including testing or a study session focused on the study content without testing? A surprising proportion avoid the inclusion of some form of testing as part of their study sessions.

The perception of effort required and perceived effectiveness

Retrieval practice has attracted a great deal of recent emphasis and again the question is why learners underutilize tasks that involve retrieval. One hypothesis relates to how learners interpret the difficulty of different study methods. Here is how this has been investigated. 

Participants are involved in several studies in which they rate two study strategies (i.e., study involving or not involving testing) for the mental effort required and the perceived effectiveness (Kirk-Johnson and colleagues, 2019). It appears the more learners perceive a study strategy as mentally demanding the less effective they perceive it to be and the less likely they would select it for future use. The researchers described this as the misinterpreted-effort hypothesis.

How do learners interpret testing?

Morehead and colleagues (2015) investigated knowledge of study strategies by asking both educators and college students about different methods. The researchers asked college students how frequently they used a list of study methods including rereading, flashcards, testing, studying with friends, reworking your notes, etc., and in a second study they presented students with scenarios in which two study options were listed and asked the participants to select which they would most likely apply. When the researchers focused on teachers, they asked about the value of different study methods. These methods determined that both college students and instructors recognized testing as productive, but follow-up questions indicated the value was perceived to be the identification of which topics still needed more work rather than recognizing that the experience of trying to recall information to answer test questions results in better understanding and retention. Of course, the identification of what to focus on is important to be efficient in the use of time, but this perspective underestimates the impact of retrieval practice as an efficient learning experience.

What to make of all of this?

It seems important to me that more advanced students appreciate the value of testing even if they focus on a different and valid explanation for why testing experiences are valuable. Effective metacognition partly depends on personal insights into how learning works. It seems more obvious that testing identifies failures of understanding than that testing builds retrieval strength so recognizing the direct benefits of retrieval practice is much less likely to be apparent from daily life experiences. Educators are in the best position to address this situation. While present understanding of the values of testing reflects a limited perspective on the part of educators, the formal training of educators should address this issue and educators should then be in an ideal position to include retrieval practice tasks as they interact with students and to encourage students to make greater use of retrieval practice as part of the time they devote to self-directed study.

The message that “what doesn’t kill you makes you stronger” is a popular way to counter difficult things in other parts of our lives and perhaps a variant of the advice can be promoted when it comes to how we try to learn. The academics have been pushing what they describe as “desirable difficulty” as a more cognitively-based interpretation of this principle. Difficulty is not an end in itself, but in situations in which tasks require certain important cognitive activity the engagement of these activities may be perceived as more demanding than not engaging such behaviors. The need is to interpret this effort appropriately.

References:

Kirk-Johnson, A., Galla, B. M., & Fraundorf, S. H. (2019). Perceiving effort as poor learning: The misinterpreted-effort hypothesis of how experienced effort and perceived learning relate to study strategy choice. Cognitive Psychology, 115,101237. http://dx.doi.org/10.1016/j.cogpsych.2019.101237

Kornell, N., & Bjork, R. A. (2007). The promise and perils of self-regulated study. Psychonomic Bulletin & Review, 14, 219 –224. http://dx.doi.org/10.3758/BF03194055

Morehead, K., Rhodes, M. G., & DeLozier, S. (2016). Instructor and student knowledge of study strategies. Memory, 24, 257–271. http://dx.doi.org/10.1080/09658211.2014.1001992

Willingham, D. T. (2023). Outsmart your brain: Why learning is hard and how you can make it easy. Simon and Schuster.

Yang, C., Luo, L., Vadillo, M. A., Yu, R., & Shanks, D. R. (2021). Testing (quizzing) boosts classroom learning: A systematic and meta-analytic review. Psychological Bulletin, 147(4), 399-435.

It is ironic that despite reading and writing on Medium for several years now, my notes and highlights from Medium were the last source I was able to bring into my personal knowledge management system. As I have explained in a previous post, I want the long-term storage and organization of my reading to end up in Obsidian, and Readwise plays an important in acting as an intermediary to that goal. Readwise can be used to automatically collect annotations and highlights from certain sources (e.g., Kindle books) and then automatically export this content to Obsidian. Recently, these processes were enabled for Medium.

The process of moving highlights and annotations from Medium to Readwise is relatively easy. The import from Medium option now appears among the Readwise options (first image). 

To export automatically, there is an export option for Readwise that sends content to Obsidian. Which of these options you use may depend on what you want to store in Obisidian. If you read a large variety of stories in Medium, you may have a specific focus for Obsidian and prefer to select just that content from Readwise that is consistent with this focus. 

Suppose you want to export the additions from a specific imported source. In that case, the stored content in Readwise will appear within the list of imported articles and you then use the inverted caret to reveal the options and select “Export highlights”. A file containing the highlights and notes will be downloaded as a file to your device and you can open this file and add to Obsidian or any other storage tool.

One more comment about the material you have stored in Readwise. Both Readwise and Medium are online tools. You can move from a stored item in Readwise back to Medium. To do this, use the inverted caret that appears at the top corner of a specific item in Readwise and select the “View in Medium” option.

AI is so new the general public does not know what to make of it yet. AI did show up for the general public after I retired. I worked as an academic, an educational psychologist, teaching and conducting research. I no longer have to evaluate assignments and wonder if students had written them on their own or was the work I evaluated written by AI. We were more concerned with whether students had paid to have written work written for them and never really figured out how to make that call. 

Certainly, AI represents a very interesting development for those who now do the work I did. It is not just evaluating the work students submit, educational psychologists consider how important cognitive skills are developed. With AI readily available, tasks assigned to hone skills and time demands must be applied without teacher supervision and tolerated because of the work and frustration involved. Writing makes a great example. Writing activities are assigned partly to help students learn to write and also to organize, evaluate, and communicate. Writing is a convenient way to ask learners to process raw information.

AI provides a dilemma. It offers a tool we all should learn to use, but requires discipline that it not be used as an alternative to personal work necessary to develop knowledge and skills. Certain distinctions must be made by educators, but also by learners and those who can influence the behavior of learners. I think of the distinctions in this way. When does AI augment personal skills, when can it productively replace personal skills, and when does the use of AI diminish the development of skills. I have written previous posts that can be organized using this classification system.  

AI and the Creative Process. 

I have come to think of myself as a creative. This is a new self-definition and I think the label has only been used lately as well. There is a certain creativity involved in conducting research and speaking or writing to inform others. Britannica describes creativity in this way – the ability to make or otherwise bring into existence something new, whether a new solution to a problem, a new method or device, or a new artistic object or form. With academic research, very little involves pure replication so the point is to come up with a new way to test an old hypothesis or a new hypothesis and a way to test it. With teaching, ways of explaining and directing individuals often require making adjustments. No, we don’t usually use the same yellowed note cards for years at a time. 

AI and Creative Information Processing

Since I have retired, I no longer conduct experiments or face students in a classroom setting. You are reading the output of what I do now. I try to process the existing research and writing of other scholars and offer my own take on this content and how it might be applied. I have mostly focused on personal knowledge management (PKM) with a focus on the processing of digital written inputs often with the assistance of AI tools. To some extent, it is the reflection on this process as I engage in it that surfaces insights I can now write about. 

Here is how I think of the roles AI might play in my creative process. I have read hundreds of books and far more journal articles and taken notes and highlighted as I read. These notes and highlights have been stored in digital format using several online services (Mem.AI, Readwise) or my own hardware (Obisidian). I can use AI to interact with my content or with the general knowledge the AI system has been trained on. 

Perhaps my interactions might be described in terms of 

1. Retrieve – I know I read something that would be useful in what I am trying to write, but I can’t remember it. Find XXX for me.
2. Explore – I have read lots of stuff and now I am trying to write about YYY. Are there things similar to XXX in what I have read I should consider?
3. Expand – How would the AI write something in a given structure?

I write what you read myself, but these different uses of AI sometimes provide a basis for what I say. With retrieve and explore, I have previously written or recorded notes from which I can construct something new. With expand, I am not building directly on what I have saved but am asking for a product I can take ideas from in a similar way that I took ideas from a book or journal article. Retrieve and Explore make use of AI directed at my content – RAG retrieval-augmented generation. Expand is based on the AI responding to a prompt without my provision of content. 

What I am describing as Retrieve and Explore may be pretty abstract. Here ere are a couple of specific example based on the Smart Connections extension to Obsidian and Mem.AI. Both have what might be described as a “related to” capability. Perhaps it is fair to describe this as an enhanced search. 

Mem.AI

The first image shows a typical note I have stored in Mem. The button within the red square asks the AI to generate “related to” notes. The second image shows the output which consists of links to other notes. The top group of note links are related based on the content of the note. The lower groups of links are based on any tags I have added. 

Obsidian with Smart Connections

Smart Connections is an AI extension for Obsidian that does several things. It can be used to input an AI prompt that returns a response “based on my notes”. The other major capability is to identify connections based on a selected note. In the main panel, you see the note I have selected. In the right-hand panel, you see the connections. When I roll the mouse over one of these links I am provided a preview of the related note. 

I use these capabilities in two ways. I identify related notes as I write to do what I have described as Retrieve and Explore. I also use these capabilities in a long-term way as I spend time within what has been popularly described as my Second Brain. If I am just spending time looking through my personal collection of notes, I can add new tags and links based on what these AI systems bring to my attention. 

Summary

AI capabilities can be used in so many ways. What I have attempted to describe here is a use that allows me to identify connections among ideas I may not have been able to generate on my own that I can then use to generate written products with more detail and depth. I have created the building blocks and the AI is augmenting my ability to put these ideas together. 

I recently wrote a post describing Twitter Chats and how the concept of using a microblog as a chat platform originally making use of Twitter had seemingly been migrated to BlueSky. For BlueSky users I realized I had left out the description of a tool that makes chats reasonably efficient.

With Twitter, there was a tool called TweetDeck that served this same purpose. What TweetDeck allowed was the creation of a multi-column display that allowed a defined subset of Tweets to be displayed in an individual column. For those participating in a chat, any contributions were to include a tag (e.g., #grabechat) in tweets and these tweets in TweetDeck would then be separated to appear in the column focused on this hashtag. You did not have to follow those individuals participating in the chat for this to work as it was basically a search function AND the search was automatically repeated at short intervals so you also did not have keep entering a search prompt during a chat session. TweetDeck was not created to serve this specific purpose, but it worked beautifully.

Tweetdeck has had a rocky history. It was free, it was blocked, and it has since emerged as a paid app.

A similar product, Deck.Blue, has emerged for BlueSky and for those interested in a chat activity, it works similarly. This is a computer app as it requires a larger screen to display multiple columns. The following two images show what a multi-column display looks like. The first two columns are the standard feeds and the third is a column I created to follow #edchat. 

The second image shows how to create a new column. The easiest way is to conduct a search for a tag and the immediate results that are then displayed. 

Deck.blue is currently available at no cost but if you intend to use this on a regular basis there is a Patreon link to support developers. I would anticipate there will eventually be a subscription fee. 

I abandoned my Twitter and TruthSocial accounts following the election. I had purposefully spent time on these sites because I found both contained so much content I knew to be factually inaccurate I felt a responsibility to respond. My professional interests include what is described as argumentation and most folks probably recognize as debate. Interactions that use facts to explore important issues are valuable, but falsehoods purposefully promoted are a very different matter and this seemed to be so common on these platforms. Now the election has passed. 

I think I can identify motives when challenging falsehoods. It is very common to experience a personal attack in reply. For some reason, my personality protects me from such attacks. I don’t lose sleep and I don’t become preoccupied. I set aside a certain amount of time a couple of times a day to check my accounts and I try to apply logic and facts.

I spent time on TruthSocial is a sort of personal experiment. I was mostly curious. Twitter was different. I joined Twitter in 2006 when it was very different. At first it was kind of silly. People posted what they had for breakfast and such. Eventually, I thought it became useful. I used it primarily as a discovery tool. Folks couldn’t really say much so the content itself was not that informative, but I found the recommendations and links to be useful. With Musk, things have changed. The content from those I followed remained constant, but the discovery function changed. This algorithm driven option definitely did not provide information “For me”. Musk’s comments always seemed to appear and he was someone I did not follow and I certainly did not find his comments informative. My strategy of challenging misinformation resulted in an increase in the same type of false claims. The algorithm appeared to interpret my activity as a sign of interest, but increasing the proportion of false claims seems very different from increasing the frequency of comments on topics of interest.

The algorithm was not the primary reason I left Twitter. Musk made a decision that Twitter was a content site, but not a site for sharing links. This was clearly not the way I thought about the opportunities Twitter provided with its limited character limit. By using the algorithm to decrease the likelihood tweets with links were seen, Musk had eliminated what I thought was the primary value of the platform. I valued links for discovery and I used links when arguing facts. Positions that refer to external sources as justification are very different to me than personal opinions. So, after the election I decided my time could be better spent on other activities.

I am starting to have second thoughts. If many people decide it is not worth their time to interact and challenge, the lies and misinformation will only increase the impact of the echo chambers. Twitter and now Facebook and Instagram have made it clear they will not try to enforce a standard of truthfulness. All of this is now argued to be a matter of free speech. I suppose this is one way to try to distance the company from the complaints of politicians, but the reality is that once algorithms are used to guide users,  companies have made decisions about what user experiences will be. Many users don’t understand and they have no way of actually knowing how their experiences are being shaped by the decisions social media companies have made in prioritizing what they see. 

The reality is that voting patterns can be predicted from differences in whether voters pursue news content and the sources voters regard as providing any news they consume. It is not the difference between FOX and MSNBC viewers I am describing here. It is the difference between those who rely heavily on social media and newspaper or television news programs. I am a Democrat and I know that those who primarily rely on social media to inform their voting decisions disproportionally vote Republican. The situation seems clearly defined. Expecting those satisfied with the status quo to seek out more diversified and quality sources of information seems naive. Social media companies have signaled they are willing to accept the “lies don’t matter” definition of free speech. I personally see Twitter as a lost cause because of the bias of the algorithm, but I can still use Facebook and Instagram to challenge falsehoods with facts and reach a more general audience. I do encourage others to explore Mastodon and BlueSky, but I am not convinced that these sources offer the access to the mixed audience necessary for political influence.