grabe

RAG and Source Grounding With a Second Brain

As we explore meaningful ways in which we can make use of the AI tools available to us at any given time, we must be aware of the limitations of AI and what the consequences of promoting the products we generate might be. My personal use of AI is to support the writing I do. This is what you are experiencing here. Misrepresentations would be embarrassing, but I write blog posts so the consequences at worst are not catastrophic.

I would not use AI to generate blog posts by entering a series of prompts to shape a product. What I want AI to do is to help me organize what I have learned and lessen the time spent translating the information I have collected into an external representation of my thinking. I use an approach that first requires me to read, highlight, and annotate sources (typically books and journal articles). Next, I export the residue of this process (the highlights and notes) into a tool that allows me to rework and connect these elements of information. Finally, I focused an AI tool on this content and prompted the tool to create a first draft product that fits some of the goals I have in mind. I use a series of technology tools in this process and I will identify several options for each stage I go through as this post unfolds. First, a couple of additional important concepts.

What I intend is a process focused on my personal decision-making and content vetting. As much as possible, I want to avoid AI hallucinations and I want to take advantage of my thinking and experiences while reducing the labor required to generate the words to externalize my thoughts. I also want to draw on concepts I might not recall at the moment of content creation but that take advantage of thinking I did at another point in time.

A couple of AI-related concepts that are important are Retrieval-Augmented Generation (RAG) and Source Grounding. RAG is a way of using AI to prioritize specified designated content in contrast to the massive summary of information used to train an AI model. I think of the goal in this way (my interpretation). I want an AI tool to use its language and communication capabilities to respond to my prompts. When possible, I want the tool to respond to my prompts using just the body of information I have provided. RAG is useful in several ways. Models are fixed at a specific point in time and new information may be available. The information used to train the model may be flawed or inaccurate relevant to a specific context of application and information can be prioritized to avoid the use of such an idea. My intent is related to this issue. I want the content produced to represent information and positions I support.

Source grounding is a similar issue but approached from the opposite direction. What sources were used by the AI tool to generate a specific part of the output? If you think in terms of notes and highlights, what was the section of highlights or the note associated with a given statement in the output?

This combination while not perfectly implemented in most tools, offers a reasonable way for an “author” to vet content they intend to take responsibility for sharing.

Specific tools and processes

Source material processing. Most of the information I write about I first encountered as a Kindle book or a journal article available to me as a PDF. I use tools to read and think about this material that allow highlighting and annotation and then the exporting of these notes and highlights containing metadata allowing identification of source and location from which the items were taken. Quantity is important here. I have a personal library of more than a hundred Kindle books and hundreds of journal articles. Kindle devices or apps allow both annotation and highlighting. I have used several pdf tools to do the same thing. Highlights is my present favorite. Again, what is essential is the opportunity to export the notes and highlights as a document I can then import for the next stage of the sequence.

Storage, post-processing, and retrieval. The popular term for describing the next stage of my process might be Second Brain. This is the notion that adults outside of a school environment have long-term learning needs – they learn something at one point in time and they could apply what they once learned after a considerable delay. Without external storage, what they once understood could easily be forgotten. What they once understood in a different context might still be useful if reworked for a different situation. Digital storage allows an external representation of what was once understood and encourages review activities encouraging new insights. Digital storage allows various approaches to search that can surface forgotten insights.

I have written frequently about the implementation of this concept. To keep things simple for this post, Obsidian is my tool of choice within this category. I can generate notes to be stored by this tool as I read. I can also copy and paste notes and highlights from other tools. ReadWise is a utility that makes this process even easier. It automatically downloads highlights and annotations I create while reading using Kindle and with an Obsidian plugin automatically checks for new highlights or notes and uploads these highlights and annotations to Obsidian. So, overtime my Second Brain grows both manually and automatically as I read.

3. Discovery, exploration, and first drafts. This is the point at which AI makes an appearance in how I get from reading and thinking to writing. Typically, second-brain advocates promote strategic exploration of second-brain content as the way to get to externalization through writing. Review your notes. Search for connections among your notes. Enter new notes based on these processes of review and connection. AI applications with RAG capabilities allow a user to explore the content they have identified in a different way. You can interact with this content directly and in a speculative way. You can propose requests for analysis or comparison. In my case, some of these highlights and notes are now at least five or more years old so sometimes I rediscover ideas that would not come to mind without this assistance. Tools that offer source grounding allow easy access to the specific pdfs and comments associated with a claim generated by the AI tool and make it easy to review the context and related arguments from the original sources. You can return to specific sections of material and consider it from a new perspective.

My hope is that the structure I offer here will encourage you to spend some time with the tools I identify. I have used the following AI tools and propose they are worth a try.

Smart Connections – an AI plugin for Obsidian

Copilot – used as an AI plugin with Obsidian

MEM.ai – an online note-taking and exploration tool with built-in AI capabilities

NotebookLM – an online note-taking and exploration tool with built-in AI capabilities

These tools are different in several ways that may be of interest. While Obsidian is an open source tool and so are the two plugins I have mentioned, the plugins require an OpenAI api that requires a subscription. I have found the use of this API is very inexpensive and far more so than the $20 a month plan for say ChatGPT. Obsidian is resident on your device. MEM.ai is an online service with a subscription price of a little over $8 a month when committing to the yearly plan. NotebookLM offers the easiest approach for source grounding and the other tools require playing with the prompt to get note titles and links. The automatic transfer of Kindle notes and highlights to the note-taking tool via Readwise only works for Obsidian. So, these AI tools will do similar things, but work in different ways with different financial costs.

Comparison with example

An example may make what I have described more concrete. Recently, I wrote a post comparing two books – The Anxious Generation and Unlocked – that offer very different perspectives on the dangers of screen time and social media to children and adolescents. I read both books generating highlights and notes and then wanted to write a post summarizing each and comparing the perspectives provided by he author. This post can be found on my blog.

I have since made use of the four AI tools I list here to generate a short first draft on this same topic. In each case, I used the following prompt:

Using my notes, write a 600 word comparison of how Haidt and Etchells describe the relationship between adolescent phone use and mental health.

I include the output from the four tools here. They are similar, but for sake of completeness all have been provided should someone want to compare and contrast the products. Of course, such comparisons are inexact as the same tool would produce slightly different outputs should the tool respond to the same prompt several times. I don’t really expect anyone to read the multiple versions of output, but recommend you read the prefacing description of the tool to determine if that tool might fit your circumstances.

NotebookLM

NotebookLM allows me to most easily demonstrate the combination of RAG and Source Grounding. The challenge within my workflow is moving content from Obsidian to NotebookLM. I have found a way to do this efficiently and explain this process in an earlier post.

If you are unfamiliar with NotebookLM, the image that follows offers a peek at the interface. NotebookLM is a web-based tool so what is displayed is what would be seen within your browser. I have started a notebook for the specific goal of comparing my notes on these two books and I have already uploaded my collection of notes and highlights into NotebookLM. The left-hand column indicates that four sources have been added. The text box for entering a prompt to “chat” with these resources appears at the bottom of this image.

In response to the prompt, the AI capabilities of NotebookLM generates a response. The following image is complicated but shows multiple things. There is the response to the prompt which I will provide following the image. Within this content, you should see small numbers associated with different sections of the output text. These numbers when selected open a window showing the segment of the source content that was used by the AI system to produce the output. The connection of something specific from the source material with a claim made in the output is what I mean by source grounding. As you can see, I then have access to the source content and can explore the context of the designated material if I want to consider whether I agree with the interpretation made by the AI.

NotebookLM prompt response

Differing Views on Phones and Mental Health

Jonathan Haidt and Pete Etchells both examine the relationship between screen time and adolescent mental health. Haidt, author of The Anxious Generation and The Anxious Generation 2, takes a strong stance, arguing that smartphones are detrimental to the mental health of young people. Conversely, Etchells, author of Unlocked: The Real Science of Screen Time, believes that current research does not support this view, suggesting instead that the relationship is more complex and nuanced than commonly portrayed.

Haidt contends that the introduction of smartphones, coupled with an increasingly safety-focused upbringing, has led to a decline in the mental well-being of Gen Z. He argues that several key factors contribute to this decline:

Social Deprivation: Smartphones supplant face-to-face interaction, leading to a lack of vital social skills development.
Sleep Deprivation: Smartphone use often extends late into the night, leading to insufficient sleep, which is crucial for adolescent development.
Attention Fragmentation: The constant stimulation from smartphones makes it difficult for adolescents to focus on tasks, potentially hindering academic performance and emotional regulation.
Addiction: The design of social media platforms exploits psychological vulnerabilities, leading to addictive patterns of use.

Haidt supports his argument by highlighting the correlation between the rise of smartphone use and the increase in mental health issues among adolescents. He points to 2012, the year when Facebook acquired Instagram and rates of depressive episodes began to rise, as a pivotal moment in this trend. He emphasizes the importance of face-to-face, synchronous interaction for healthy emotional development, a type of interaction that social media platforms largely fail to provide. Instead, social media, with its emphasis on likes, shares, and followers, fosters conformity and potentially harmful social comparisons. Haidt is especially concerned about the impact of social media on girls, noting that the worst years for social media use for girls are between 11 and 13. He advocates for delaying smartphone ownership until high school and restricting social media use until age 16. He also argues for phone-free schools and more opportunities for unsupervised play, allowing children to take risks and develop resilience.

Etchells, however, cautions against drawing simplistic conclusions about the relationship between screen time and mental health. He argues that many studies linking smartphone use to negative outcomes rely on self-reported data, which can be unreliable and influenced by prevailing societal narratives about screen time. He points out that large-scale, longitudinal studies that track actual smartphone usage over extended periods are scarce, making it difficult to establish causal relationships. He also emphasizes the importance of considering the context of smartphone use, rather than focusing solely on the amount of time spent online. Factors such as the specific apps used, the purpose of use, and the social context of online interactions can significantly influence the impact of screen time on mental well-being.

Etchells critiques the use of anecdotal evidence and alarmist headlines in the media, arguing that such stories, while compelling, can distort public understanding of the issue. He suggests that focusing on developing healthy screen habits, rather than demonizing technology, is a more productive approach. He advocates for a nuanced understanding of screen time, acknowledging both its potential benefits and risks.

Both Haidt and Etchells agree that more research is needed to fully understand the complex interplay between screen time and adolescent mental health. While their perspectives differ in terms of the severity of the issue and the proposed solutions, both authors offer valuable insights into a topic of increasing societal concern.

Smart Connections and Copilot

Smart Connections and Copilot are both plugins added to Obsidian and both rely on the OpenAI API (requires a subscription). It makes sense to simply suggest these plugins are options that do very similar things.

If you are unfamiliar with Obsidian, the following image is intended to show the basic functions. The left-hand (note folders) and middle column (a selected note) concern displaying, locating, and inputting notes. The right-hand column serves a variety of additional functions including accessing and using plugins. Two symbols in the red square at the top of this column allow the selection of either the Copilot or Smart Connections capabilities. The text box within the red square at the bottom is for the submission of a prompt to the AI plugin. The response to the prompt appears in the now vacant area above the prompt window.

Smart Connections Prompt Response

Based on your notes, Jonathan Haidt and Pete Etchells offer contrasting perspectives on the relationship between adolescent phone use and mental health. Haidt, in his book “The Anxious Generation,” argues that smartphones and social media have contributed to a surge in anxiety and depression among young people, particularly focusing on the impact of the widespread adoption of technology in the late 2000s and early 2010s. He proposes solutions such as delaying smartphone and social media access until high school, implementing phone-free schools, and promoting unsupervised play to address the mental health challenges faced by adolescents.

On the other hand, Etchells, in his book “Unlocked,” challenges the common narrative surrounding screen time and mental health. He argues that the issue is more complex than simply quantifying screen time and emphasizes the importance of how individuals engage with technology. Etchells suggests that some screen time can be beneficial and urges readers to reflect on their screen habits, considering factors such as content consumed, apps used, and context of use. He encourages a more thoughtful and intentional approach to technology use to support and improve mental health.

Haidt’s perspective is supported by research indicating a negative correlation between social media use and life satisfaction, particularly among younger adolescents. He highlights key dates associated with changes in technology, such as the introduction of iPhones in 2009 and the acquisition of Instagram by Facebook in 2012, correlating with a significant upturn in depressive episodes among Gen Z around 2012. Haidt’s argument is further reinforced by the addictiveness of social media, which exploits users’ identities and impacts emotional development.

In contrast, Etchells challenges the notion that screen time is a key driving factor behind the decline in mental health. He emphasizes the need to focus on how individuals engage with technology rather than solely measuring screen time. Etchells suggests that understanding and reframing screen habits, reflecting on behaviors, and being critical of information can help individuals leverage technology to support their mental well-being.

Overall, while Haidt highlights the detrimental effects of smartphone and social media use on adolescent mental health, Etchells offers a more nuanced perspective, advocating for a shift in focus towards healthy screen habits and intentional technology use. Both authors contribute valuable insights to the ongoing debate on the relationship between adolescent phone use and mental health, providing readers with different perspectives to consider when addressing this complex issue.

CoPilot Prompt Response

To write a comprehensive comparison of how Jonathan Haidt and Tim Etchells describe the relationship between adolescent phone use and mental health, we need to delve into their respective perspectives and analyses. Both authors have explored this topic, albeit from different angles and with varying emphases

Jonathan Haidt’s Perspective

Jonathan Haidt, a social psychologist, has extensively researched the impact of social media and smartphone use on adolescent mental health. Haidt argues that there is a significant correlation between increased smartphone use and the rise in mental health issues among adolescents, particularly anxiety and depression. He often cites statistical data and longitudinal studies to support his claims, pointing out that the surge in mental health problems coincides with the widespread adoption of smartphones around 2012.

Haidt emphasizes the role of social media platforms in exacerbating feelings of inadequacy and social comparison among teenagers. He suggests that the curated nature of social media content leads adolescents to constantly compare themselves to others, which can result in negative self-perception and increased anxiety. Furthermore, Haidt highlights the addictive nature of these platforms, which are designed to capture and hold users’ attention, often at the expense of real-world interactions and sleep, both crucial for mental well-being.

Tim Etchells’ Perspective

Tim Etchells, on the other hand, approaches the topic from a more nuanced and perhaps less deterministic angle. While acknowledging the potential negative impacts of smartphone use, Etchells is more focused on the context and manner in which these devices are used. He argues that smartphones are not inherently detrimental to mental health; rather, it is the way they are integrated into daily life that matters.

Etchells emphasizes the importance of digital literacy and the need for adolescents to develop healthy habits around technology use. He suggests that when used mindfully, smartphones can offer significant benefits, such as fostering connections, providing access to supportive communities, and serving as tools for creativity and self-expression. Etchells is more likely to highlight the potential for positive outcomes, such as increased access to mental health resources and the ability to maintain relationships across distances.

Comparative Analysis

When comparing Haidt and Etchells, several key differences emerge in their analyses of the relationship between adolescent phone use and mental health. Haidt tends to focus on the negative consequences, often presenting a more cautionary tale about the pervasive influence of smartphones and social media. His approach is data-driven, relying heavily on empirical evidence to underscore the risks associated with excessive use.

In contrast, Etchells offers a more balanced view, acknowledging both the potential harms and benefits of smartphone use. He advocates for a more individualized approach, considering the diverse ways in which adolescents engage with technology. Etchells is more optimistic about the potential for positive outcomes, provided that there is an emphasis on education and the development of healthy digital habits.

Both authors agree that there is a relationship between smartphone use and mental health, but they diverge in their interpretations and proposed solutions. Haidt calls for more stringent regulations and societal changes to mitigate the negative impacts, while Etchells focuses on empowering individuals through education and awareness.

Conclusion

In summary, Jonathan Haidt and Tim Etchells provide contrasting yet complementary perspectives on the relationship between adolescent phone use and mental health. Haidt’s analysis is rooted in caution and concern, highlighting the potential dangers and advocating for systemic changes. Etchells, meanwhile, offers a more hopeful outlook, emphasizing the role of education and personal responsibility in navigating the digital landscape. Together, their insights contribute to a more comprehensive understanding of this complex issue, underscoring the need for a balanced approach that considers both risks and opportunities.

MEM.ai

I started Mem.ai because it was from the beginning a note system that incorporated AI. I had to manually upload the notes I entered in Obsidian. Mem.ai proposed that the linking and tags advocated as so important in Obsidian were less necessary when one could apply AI capabilities to explore the stored notes. There seemed some logic to this proposal and I wanted to see what I thought after using both systems.

MEM.ai Prompt Response

Certainly! Let’s delve into a comparison of how Jonathan Haidt and Pete Etchells describe the relationship between adolescent phone use and mental health, using insights from your notes.

Comparing Perspectives: Haidt vs. Etchells on Adolescent Phone Use and Mental Health

The discourse surrounding adolescent phone use and its impact on mental health is a complex and multifaceted one. Jonathan Haidt and Pete Etchells offer distinct perspectives on this issue, each bringing unique insights into how digital technology influences the well-being of young people. While Haidt presents a more cautionary view, Etchells provides a nuanced analysis that considers both risks and benefits.

Haidt’s Perspective: A Warning on Digital Dangers

Jonathan Haidt, a prominent social psychologist, has been vocal about the potential dangers of smartphone use among adolescents. His perspective is largely shaped by the following points:

1. Correlation with Mental Health Decline: Haidt highlights a significant uptick in mental health issues, such as anxiety and depression, among teenagers starting around 2011. He correlates this trend with the rise of smartphones and social media, suggesting a causal link between increased screen time and deteriorating mental health.

2. Social Media’s Role in Psychological Distress: Haidt argues that social media platforms exacerbate feelings of inadequacy and low self-esteem, particularly among adolescent girls. The constant exposure to curated images and the pressure to gain social validation can lead to harmful comparisons and social anxiety.

3. Gender-Specific Vulnerabilities: Haidt emphasizes that adolescent girls are especially vulnerable to the negative effects of social media. He attributes this to the heightened focus on appearance and social comparison, which are more pronounced in female social media interactions.

4. Advocacy for Restrictive Measures: To combat these issues, Haidt advocates for delaying smartphone use among children. He suggests policies like “No smartphones before high school” to reduce exposure to potentially harmful digital environments and promote healthier developmental outcomes.

Etchells’ Perspective: A Balanced Approach

In contrast, Pete Etchells offers a more balanced view of the relationship between adolescent phone use and mental health. His analysis considers both the potential risks and the benefits of digital technology:

1. Empowerment through Connectivity: Etchells acknowledges that smartphones and social media can empower adolescents by providing access to information, educational resources, and supportive communities. This connectivity can foster a sense of belonging and enhance self-efficacy.

2. Importance of Digital Literacy: Etchells stresses the role of digital literacy in helping adolescents navigate the online world safely. By equipping young people with critical thinking skills and the ability to manage their digital presence, they can mitigate potential negative impacts.

3. Balanced Screen Time: Rather than advocating for strict limitations, Etchells promotes a balanced approach to screen time. He suggests that integrating both digital and physical activities can help adolescents develop a healthy relationship with technology.

4. Focus on Resilience and Adaptation: Etchells believes that with proper guidance, adolescents can develop the resilience needed to thrive in a digital world. By fostering emotional intelligence and critical thinking, young people can learn to navigate the challenges of social media and smartphone use.

Contrasting Approaches to Solutions

While both Haidt and Etchells recognize the significant impact of smartphone use on adolescent mental health, their approaches to addressing these challenges differ:

Restrictive vs. Empowering Solutions: Haidt’s approach is more restrictive, focusing on limiting exposure to technology as a means of protecting mental health. In contrast, Etchells emphasizes empowerment through education and digital literacy, advocating for a more nuanced understanding of technology’s role in adolescents’ lives.

Focus on Risks vs. Opportunities: Haidt’s narrative centers around the risks and negative outcomes associated with smartphone use, particularly for adolescent girls. Etchells, however, highlights the opportunities for empowerment and connection that technology can provide, suggesting that the benefits can outweigh the risks when managed appropriately.

Conclusion: Navigating the Digital Landscape

The contrasting perspectives of Haidt and Etchells underscore the complexity of the relationship between adolescent phone use and mental health. While Haidt’s concerns about the negative impacts are valid, Etchells’ emphasis on empowerment and education offers a hopeful path forward. Ultimately, the key lies in finding a balance that maximizes the benefits of technology while mitigating its risks, equipping young people with the tools they need to thrive in an increasingly connected world.

This comparison captures the essence of both authors’ views, providing a comprehensive look at their differing approaches to adolescent phone use and mental health. If there are specific points you’d like to adjust or expand upon, feel free to let me know!

—-

My goal is to offer some options you might consider. I find advantages and disadvantages with each approach and still use a couple of these tools when I think the first one could be improved.

Damage from time online?

The benefits and harms done by our devices and the ways we use them is one of those topics likely of at least marginal interest to most people. We wonder whether we have become too dependent and whether existing interests and capabilities have deteriorated. If we are a parent or educator we likely have additional unique interests. Have I messed up my kid by letting him or her use a phone without supervision? How do I keep my students on track in the classroom and are they using AI to complete my assignments quickly without any actual educational benefits?

My personal interest goes a bit further in that I spent my career preparing educators to use technology in their classrooms and I by training feel responsible for understanding the science behind recommendations made to others. So, I review the research on many topics. From this perspective, I would offer the following observation – it is seldom as simple as it seems. I know this to be true in the social sciences which would be the field covering phone use and the impact on development and learning. This is the case for many reasons and far from a topic I can thoroughly consider here. One comment I used to make to my own students may be helpful. For context, I was trained as a biologist who became a psychologist. My comment was – students tend to see a discipline like chemistry as more sophisticated and complicated than psychology. Consider that the chemicals in that beaker don’t decide whether they feel like reacting today. People are different.

Back to the issue of our devices and the research on how we are being impacted. I recently read a book by Jonathan Haidt titled The Anxious Generation. Many people must have read this book because it topped the New York Times best-seller list for several weeks which I interpret to mean many parents and educators are concerned and were attracted by the topic. Recommendations made in the book, for example, no phones in schools, have been implemented in many schools. I read Haidt’s book in response to such changes concerned my pro-technology advocacy should be tempered. I admit my initial reaction was that Haidt made a reasoned and evidence-based case. However, I have learned that such secondary sources need to be vetted and I was concerned by some of Haidt’s rationale which I would describe as “what else besides iPhones and Instagram” could explain the mental health issues of adolescent females. The “what else could it be” logic just seems weak. In reaction, I read a related, but less well-known book by Pete Etchells and as I suspected the issue is complicated.

What follows is my summary of the two books with some related comments.

Comparing Perspectives on Screen Time and Mental Health

Haidt’s “The Anxious Generation”

In “The Anxious Generation,” Jonathan Haidt really covers two broad problems he contends have changed the mental health development of children and adolescents. Children have been overprotected in their daily lives resulting in a lack of play and autonomy involving risks and consequences. In contrast, the author argues that the rise of smartphones and social media have been largely unmonitored and unrestricted with negative consequences to mental health. I will comment on both claims, but emphasize technology use.

Haidt highlights the following key points:

Slow-Growth Childhood: Human children have an extended childhood compared to other mammals, providing them with time to learn through play and social interactions. This extended childhood is crucial for emotional and cognitive development.

Importance of Play: Play-based childhoods are essential for healthy development, fostering skills such as emotional regulation, social competence, and creativity.

Safetyism: Haidt argues that the overprotective parenting style prevalent in the 1990s, termed “safetyism,” limited children’s opportunities for risk-taking and independence, further contributing to anxiety. Haidt uses example older folks will find familiar – e.g., playground equipment that is no longer allowed for safety reasons and reduced independence in moving about the neighborhood or in solo trips to the store.

Negative Impacts of Smartphones: These problems include:

Social deprivation: Replacing face-to-face interaction with screen time.

Sleep deprivation: Late-night phone use cutting into sleep at ages when necessary. The early school start for adolescents is an example of when this becomes a problem.

Attention fragmentation: Constant notifications and multitasking impairing focus.

Addiction: The design of social media platforms promoting addictive behaviors. Etchells will reject this notion if understood as a physiological addition as in drug use.

Puberty as a Sensitive Period: Puberty is a critical period for brain development, making adolescents particularly vulnerable to the negative effects of social media. Hence, experiences allowed at other ages may do less damage.

What evidence suggests that puberty is a “sensitive period” for social media use?

Brain Development: Puberty to a greater degree than the earlier period of childhood involves the pruning of neurons and the strengthening of connections, processes influenced by experiences. This heightened neuroplasticity implies that experiences during puberty, including social media use, can have a greater and more enduring impact on brain structure and function.

Orben’s Research: Amy Orben, found that negative correlations between social media use and life satisfaction were more pronounced in the 10-15 age group, encompassing puberty, compared to older individuals (16-21). This finding directly supports the idea of puberty as a sensitive period for social media’s influence on well-being.

Social Learning and Conformity: Adolescents are conformers, a process that social media platforms can strongly influence. Puberty marks a time when individuals are particularly sensitive to social cues and peer influence, seeking to establish their identity and social standing. Social media platforms, with their emphasis on likes, followers, and influencers, can hijack this natural drive for social learning, potentially leading to distorted perceptions and unhealthy comparisons. Giirls are most vulnerable to social media’s negative effects between 11 and 13, while boys experience this heightened vulnerability between 14 and 15. This timeframe aligns with typical puberty onset, further suggesting that the hormonal and social changes during this period make adolescents more susceptible to the pressures and comparisons prevalent on social media.

In conclusion, the sources provide converging evidence pointing to puberty as a “sensitive period” for social media use. This sensitivity stems from the interplay of rapid brain development, heightened social awareness, and the drive for conformity, all characteristic of adolescent development. This information suggests that social media use during puberty requires particular attention and potentially different approaches compared to other age groups.

Haidt makes use of historical data linking changes in mental health data over the years to changes in the technology available to young people. Much of Haidt’s argument about the damage done by access to the Internet is made by tracking mental health outcomes against dates associated with changes in technology. Gen Z started to reach puberty in 2009. iPhones became available in 2009. Social media capabilities such as retweets and likes became available in 2009. Front facing cameras in 2010. Facebook acquired Instagram in 2012. His challenge to critics is what else could have accounted for the changes in adolescent mental health that correspond to this time frame.

Recommendations: Haidt proposes four solutions to address this issue: no smartphones before high school, no social media before age 16, phone-free schools, and promoting more unsupervised play and childhood independence.

Etchells’ “Unlocked”

Pete Etchells, in his book “Unlocked: The Real Science of Screen Time (and how to spend it better)” offers a more nuanced view of screen time and its relationship to mental health.

Etchells first notes that the consequences of technology and social media likely include both positive and negative impacts so explanatory models must take into account both possible outcomes. Etchells’ key arguments include:

Screen Time is Complex: Etchells argues simply measuring the total amount of screen time is insufficient to understand its impact on well-being. What matters is how screen time is used, the specific apps and content consumed, and the context of use.

Limited Evidence for Strong Negative Effects: While some studies have shown correlations between screen time and mental health issues, the evidence is mixed, with many studies showing small or inconsistent effects. The existence of correlations even if the correlation reflected a causal relationship with negative changes in mental health as a consequence do not necessarily show large effects. The impact is typically small.

Etchells critiques the reliance on anecdotal evidence and highlights methodological limitations in much of the research. One of the most important methodological limitations appears to be whether self-reports versus verifiable behavioral data are used. Even though digital devices are great at recording data on use (consider your iPhone and your daily report of time spent on different apps) many studies ask users to estimate their time spent and activities experienced. When studies using self-report data are compared with studies using the more accurate data collected by the devices used, significant relationships are typically found only with the self-report data. Etchells speculates that user awareness of screen time issues and potential negative consequences results in participants in these studies interpreting negative personal experiences as an indication of too much screen time. How individuals feel about their screen use and their perceived self-control are more strongly related to well-being than objective measures of screen time.

Importance of Habits over Addiction: Etchells argues that framing excessive screen use as “addiction” is unhelpful, as it focuses on abstinence as the only solution. Instead, he advocates for viewing screen use as a set of habits that can be modified to promote well-being. The author argues that the assumption of a physiological explanation (probably dopamine) within the framework of addictions such as drug abuse is what readers of many of the negative books assume. Addiction tends to ignore agency and motivation. This is where the notion of habits also fits and solutions that are more nuanced with a focus on decision-making and awareness.

Oversimplification of “Screen Time”: Etchells argues that simply measuring the total amount of time spent in front of screens is too crude a metric to be meaningful. He emphasizes that “screen time” encompasses a vast range of activities, from educational apps to social media to video games, each with potentially different effects on well-being. He contends that focusing solely on duration ignores the crucial factors of content, context, and purpose of use.

Reliance on Anecdotal Evidence: Etchells criticizes the prevalent use of anecdotal stories to support claims about the negative impacts of screen time. While these stories can be compelling and relatable, he argues they often lack scientific rigor and can lead to biased conclusions. We like stories because we often can relate to such examples and do not necessarily make the same connections with the data in graphs and statistics. Of course, coming up with examples that fit a perspective does not necessarily fit what is most common in a sample of participants. He points out that reliance on anecdotes is particularly problematic in a relatively new field like digital technology research, where robust, longitudinal data is still limited.

Methodological Issues in Correlational Studies: Much of the research on screen time relies on correlational studies, which can only demonstrate associations, not causal relationships. Etchells highlights the problem of “third variables,” unaccounted-for factors that might influence both screen time and mental health, leading to spurious correlations. For example, pre-existing mental health conditions, family dynamics, or socioeconomic factors could contribute to both increased screen time and negative well-being outcomes, creating the illusion of a direct link where none exists. Correlations showing relationships do not necessarily convey the magnitude of an effect. A significant correlation in a large population may be associated with a very small impact.

Lack of Theoretical Framework: Etchells argues that the field lacks a robust theoretical framework to guide research and interpret findings. He suggests that without clear theoretical models, researchers are left to “grasp at straws,” making tenuous connections between screen time and a wide range of outcomes without a solid foundation for understanding the underlying mechanisms. This lack of theoretical grounding makes it difficult to develop testable hypotheses and draw meaningful conclusions. Haidt’s “what else could it be” argument fits here.

Ambiguous Terminology: Etchells points out the imprecise use of terms like “addiction” and “attention” in screen time research. He criticizes the tendency to label excessive screen use as “addiction” without sufficient evidence of a true physiological dependence. He also argues that “attention” is often used interchangeably with “self-control,” leading to conceptual confusion and misinterpretations of research findings.

Insufficient Attention to Positive Effects: Etchells argues that the focus on potential negative consequences of screen time has overshadowed research on its potential benefits. He acknowledges that excessive or problematic screen use can be detrimental, but he emphasizes the importance of recognizing the many ways in which technology can enhance learning, communication, and social connection. He encourages a more balanced approach that considers both the positive and negative aspects of screen time.

In conclusion, Etchells urges caution against drawing sweeping conclusions about the impact of screen time based on the existing research. He advocates for more nuanced investigations that consider the complexities of technology use, adopt rigorous methodologies, and develop strong theoretical frameworks to guide future research.

Comparing Haidt and Etchells

While both Haidt and Etchells acknowledge the potential downsides of excessive screen time, they differ significantly in their overall perspectives and conclusions:

Emphasis on Negative Effects: Haidt emphasizes the negative impacts of smartphones and social media, attributing a rise in mental health problems directly to these technologies. Etchells, on the other hand, acknowledges the potential for harm but argues that the evidence for strong negative effects is weak and often overstated.

Role of Social Media: Haidt sees social media as a particularly harmful force, promoting conformity, comparison, and addiction. Etchells recognizes these potential issues but maintains that social media can also have positive benefits, depending on how it is used. The existence of both positive and negative outcomes argues against simplistic abstinence/avoidance and urges more careful personal awareness and control.

Solutions: Haidt proposes strict limitations on smartphone access and social media use, advocating for a return to a more play-based childhood with less adult supervision. Etchells focuses on promoting healthy screen habits, emphasizing individual agency and self-regulation rather than strict restrictions.

Conclusion

Haidt’s “The Anxious Generation” presents a compelling, albeit alarming, argument about the potential downsides of technology for young people. However, Etchells’ “Unlocked” offers a more balanced perspective, urging caution against oversimplifying the complex relationship between screen time and mental health. The key takeaway is that understanding the nuances of screen use and focusing on developing healthy habits is crucial for mitigating potential harms and maximizing the benefits of technology. Both books highlight the importance of critical thinking and evidence-based approaches when evaluating the impact of technology on our lives.

Sources:

Etchells, P. (2024). Unlocked: the real science of screen time (and how to spend it better). Hachette UK

Haidt, J. (2024). The anxious generation: How the great rewiring of childhood is causing an epidemic of mental illness. Random House.

Hooked on Online Writing

I have been reading Robert Haidt’s “The Anxious Generation”. This post is not a description of that book, but the book’s focus and content were the origin of my present topic. The book, which has topped the New York Times best-seller list for some weeks, focuses on the mental health damage of the way kids are raised in combination with the negative impact of social media. Haidt is particularly concerned about young girls and their susceptibility to the negative consequences of what becomes a damaging addiction to social media. While male and 75, I had an uneasy feeling Haidt could have been describing me and my preoccupation with a different online environment. Before I get to my personal insight, allow me to describe the characteristics Haidt argues drive the general problem that is the focus of his book.

Haidt proposes that the writings of B.J. Fogg have served as the bible guiding many social media entrepreneurs. I usually read the core literature associated with what I write, but Fogg’s book was priced at textbook levels and I decided I was very familiar with the central content (a different source). Fogg successful persuasion can be accomplished based on behavioral principles any college student who has taken the Introduction to Psychology course will recognize. His terminology is a little different. When I taught the Intro course, I described the process Fogg and Haidt emphasize as operant conditioning. A very simple version of the way operant conditioning changes behavior is captured in the sequence – stimulus – response – consequences. Behavior changes (response) are encouraged by positive consequences. When used to account for what seems an addiction to online social media in children, the response might be frequent checking of a social media account and the positive reinforcers (likes, comments, attention, etc.). Access to a phone (the stimulus) triggers the initial behavior generating the consequences.

The Fogg version is described in the following model.

Fogg uses a little different vocabulary. The external trigger is his term for stimulus and action is the behavior. Variable reinforcement is another concept from behavioral psychology that translates as a situation in which a behavior does not produce a consequence every time it is produced. This unpredictability increases the frequency of behavior. A common example is the way a slot machine works. A gambler would soon quit if previous experiences were always wins and now the machine stops paying out. You tend quit putting coins into a vending machine if the first attempt or maybe the second produces no soda or candy bar. If however, a slot machine generates wins now and then people keep feeding their coins.

Fogg and Haidt add one additional component to the model – investment. Social media often has another characteristic increasing holding power. The example of investments are everywhere online. Do you play that online game where you have to guess the spelling of a mystery word within so many tries? If you have a streak of days going, you have an investment that makes it very likely you will not miss a day. Do you take photos or use AI to create photos to embellish your posts? You are making an investment. Do you pay to add weapons or clothing (skins) in an online game? You are investing. When your identity becomes part of your participation in an activity, you are heavily invested. I would suggest making political comments on social media are a good example of being invested. Our political affiliations are part of our identities. Haidt argues that Instagram has such a powerful impact on young females (often negative) because the photo-heavy nature of the platforms triggers the role appearance has in the identity of young women.

Social media involves a switch from external triggers to internal triggers. Once you are involved, you don’t have to be sent a message that you have a new like or comment. You don’t have to see your phone sitting on your desk. Your thoughts lead you to get your phone out of your pocket or purse to check your accounts to see if anything new has shown up.

Haidt emphasizes the powerful impact of social media on the attention and mental health in children and adolescents. I saw a similarity in the application of the model and the arguments made to adults who write on blogs, Substack and Medium. I don’t think the negative impact holds in the same way because of the life experiences and brain maturity (frontal cortex and metacognition) of adults and draw the following parallels more out of amusement than concern. If there was one insight that triggered me to write this post, it was the recognition of identity in motivating behavior more so than the behavioral explanation. For those who write, writing for public consumption is part of personal identity. I think the behavioral impact on behavior is more powerful because of this self view.

Here is my attempt to apply the cycle of engagement (Haidt and Fogg) to writing:

1. Trigger

The cycle starts with a trigger that prompts users to take an action. Triggers can be:

External triggers: Notifications, emails, messages, ads, or reminders that tell users to open an app or take action. For example, a push notification letting you know someone liked your post.
Internal triggers: Emotions, thoughts, or desires that come from within, like boredom, curiosity, just wondering if there is something there. These feelings push users to check their phone or social media without an external prompt.

2. Action

After the trigger, the user performs an action. The action may be simple such as seeing if anyone has read your post or liked your content. There may be statistics or charts to check. The actions that are triggered can be more involved such as researching and writing another post.

3. Variable Reward

Once the action is taken, the user receives a reward, but to keep engagement high, the reward is often variable or unpredictable. This uncertainty makes the reward more powerful or resistant to extinction in behavioral terms.

Social rewards: Users may get more or fewer likes, comments, or shares on their posts each time, keeping them hooked. In some writing platforms, there is financial compensation to check.

4. Investment

The final stage of the cycle is investment, where the user puts something of value back into the platform. This investment can be time, effort, data, or emotional input, and it increases the likelihood that the user will return to the platform. Examples include:

Creating content: Posting a photo, video, or comment.
Personalization: Customizing a profile or favoriting the work of others.
Building relationships: Adding friends, replying to messages, or participating in communities.This investment helps users feel more connected to the platform because they’ve contributed, and it primes them for future triggers. As they invest more, they become more likely to engage again, as their investment increases the value of the platform to them
Financial commitment: Payment of a fee to participate.
Affirmation of identity: The content I generate and share demonstrates I am a writer.

5. Repeat

Once users have invested in the platform, they receive new triggers, starting the cycle all over again. Over time, this cycle builds a habit, with users increasingly relying on internal triggers, such as boredom or curiosity, to engage with the platform.

If you are a writer who contributes content online, you may see yourself in this description. With social media, such models are used to describe how participants are drawn into spending more and more time in such environments sometimes with negative emotional consequences. I will leave it to your own analysis to determine whether being drawn into a platform to which you contribute your writing impacts your emotional well-being.

Reference

Fogg, B.J. (2003). Persuasive technology: Using computers to change what we think and do (Interactive Technologies).

Upload Obsidian Folders to NotebookLM

Many of us have now had the opportunity to spend considerable time with AI tools and have sorted out for ourselves what types of applications are most productive. For my personal use, I found the greatest value in the use of AI to analyze the notes and highlights I have accumulated over the years to search for insights I might use in my writing projects. I keep most of my notes in Obsidian and one of the more popular posts I have generated focusing on AI and personal notes explained my approach to using the Obsidian AI extension Smart Connections.

As I continue exploring, I have found another approach that others may find useful. This approach uses the Google AI tool NotebookLM. This tool is presently provided at no cost and is easy to use. There is a frustrating mismatch between how most of us use Obsidian and the structure of NotebookLM. Obsidian users typically create many short notes and then create links and tags to allow these notes to be stored and used in the most flexible manner possible. NotebookLM allows the uploading of considerable content into an individual notebook to be explored with AI chats, but limits the number of files that can be included. This combination – many small files in Obsidian and a limited number of files in NotebookLM notebooks – creates a challenge.

Here is my present solution. There is an extension for Obsidian (Better Export PDF) that allows the content from multiple files stored within an Obsidian folder to be exported as a single PDF. You can probably see where this is going. Export the many files in several folders as PDFs and then add these PDFs to NotebookLM.

The Better Export PDF in Obsidian

The following image was captured as I used Obsidian. You can see the multiple folders I have created. After One Better Export has been added to Obsidian, you first right-click (control click) on a folder to bring up various options. The one you want is Export folder to PDF.

Activating this option may leave you confused as the process takes some time to accomplish. You know when things are working properly when you see the names of individual files appear towards the top of the display (red box). When the process of concatenating the files has been completed the complete PDF file will appear in what initially appears as white space. Now you export this large file.

You complete this process additional Obisian folders until you have exported the large PDFs you want to explore in NoteBookLM.

NotebookLM AI Chat

The following image shows a notebook I have created based on three PDFs (Cognitive, Generative, Notetaking). A couple of hundred individual notes are included in these three large files. At the bottom of the image, you should find a prompt I have entered.

For AI tool users, everything should now be more familiar. NotebookLM has responded to my original prompt and I can accept some of the recommendations for related topics I might explore (not shown). NotebookLM offers a way to review the content the AI used to generate the prompt reply (numbers in the red box). Selecting one of these numbers takes you to the section in your input and allows you to explore that context.

Summary

So, this combination of Obsidian and NoteBookLM seems to work well. The key is the Obsidian extension Better Export PDF which allows many Obsidian notes to be combined to address the limited number of files NoteBookLM will accept as input. Give it a try.

Reading and what we know

In reading and then writing about Willingham’s The Reading Mind, the author makes some interesting arguments about the relationships between what we know and reading comprehension and between the time spent reading and what we know. As the reciprocal relationship implies, this is a cognitive explanation of how the rich get richer. The author throws in a related analysis of whether our commitment to spending time with technology has diminished the likelihood that we are now benefiting from this relationship. What follows is my embellished summary with a few updates.

I like to think of Willingham’s description of reading as explaining how the inputs to comprehension come simultaneously from opposite directions. The bottom-up inputs come from the page – letter/phoneme recognition, word recognition, word meaning, broader understanding. The top-down inputs come from stored information (memory) – general knowledge, specific knowledge, broad understanding of the passage being read, sentence meaning, etc. You probably recognize that these inputs are really the same series just listed in opposite directions. It turns out these inputs help each other out as many are being processed at the same time. Some of these interactions you may not recognize and may surprise you. For example, a letter at the beginning of a word can be identified faster than the letter in isolation. Others, you may recognize if I bring them to your attention. For example, the understanding of a sentence may help you assign meaning to an unfamiliar word in that sentence. The battle over what is commonly described as the “science of reading” is focused on the early bottom-up processes. Should early reading experiences emphasize associating sounds with word components or should word recognition and context receive greater attention? It might be helpful to understand why the science of reading may seem to change over time because effective reading depends on multiple processes acting to support each other. Many subskills end up being important. I am more interested here in the top-down processes. Once we get past learning the basics of reading, how does reading both depend on and develop what we know?

Willingham proposes that it is never possible for a writer to explain a topic in complete detail and an author must rely on what readers already know to fill in some elements from existing knowledge. So when a reader is exposed to new material understanding is dependent both on reading skill and existing knowledge. What we already know has a surprisingly large impact on what we comprehend and retain. This claim can be demonstrated on two levels – general knowledge and topic specific knowledge.

Anne Cunningham and Keith Stanovich conducted an interesting study of high school students in which they had the students take a test of general knowledge. General knowledge involves all kinds of random information. In their research, students were asked about factual knowledge of science, history, and literature and the accomplishments of people from history, science, sports, and music. It was not that the knowledge of the specific topics was necessarily important, but rather that the scores obtained were predictive of the level of general knowledge more broadly. The researchers also administered a standard test used to evaluate reading comprehension skills. These two variables – general knowledge and comprehension skill were strongly correlated. More capable readers probably have learned more from reading, but what they have learned may also be a factor in determining their reading skill. Such differences are partly responsible for the advantage some students gain from general life experiences that cannot be provided while in the classroom.

The advantage of what we know to reading performance is made clear in studies that investigate the importance of specific knowledge in understanding content related to that knowledge. Willingham used a study based on reader differences in understanding the game of soccer. This surprised me as I was aware of a very similar study based on the game of baseball. A general description of the clever methodologies of these studies will explain how reading skills and relevant knowledge were differentiated. These studies (Recht & Leslie, 1988; Schneider, Korkel & Weinert, 1989) were conducted with younger learners. Depending on the study, these learners were given a test to evaluate their knowledge of baseball or soccer. Scores on reading comprehension tests were also available. These two measures were used to identify four groups of learners – high on both measures, high on one measure and low on the other, low on both measures. After reading a story about part of a baseball game or a soccer match, readers were asked to recall as much as they could from the story. The grouping of readers allowed a way to statistically isolate the impact of reading ability from background knowledge on what was retained. Relevant knowledge was at least as important as reading proficiency to what was taken away from the reading experience. By the way, this type of research should be relevant to those who argue access to Google is equivalent to knowing things. What you can find through search does not offer the same benefit to immediate processing as what you already know.

Leisure Reading and Technology

Leisure reading plays a unique role to developing readers because the time devoted to leisure reading varies far more than the time spent reading in schools. As adults, if we read many of us only engage in reading that would fall in this category. If engaged in an educational setting, leisure reading augments assigned reading as a reading skill development opportunity and as an opportunity to expand the acquisition of general knowledge and that is important in improving the effectiveness of reading.

In the discussion of technology and reading, Willingham considers both whether reading from a screen offers the same benefits as reading from paper and whether our constant use of technology has displaced time spent reading. Willingham acknowledges the research that demonstrates a small benefit for reading from paper. Like my posts on this topic, he sees this difference to be of little consequence acknowledging that the root cause is unknown. My embrace of screen reading is related to the long-term advantages of the production of digital notes and searchable highlights that can be organized and efficiently searched.

The notion that screen time comes at the cost of reading time (displacement) was countered with what to me were some surprising data. In his focus on leisure reading, Willingham argues that Americans never did read much and that this amount of time has not diminished from pre-internet days. I checked Willingham’s sources on this topic and found a more recent survey of adolescent reading behavior (Rideout and colleagues 2022). Reading time has actually increased a bit, but the average daily reading time has now reached 34 minutes. Twenty-four minutes involve books (paper or ebooks) and the rest newspapers, blogs, and other long-form content. I don’t find this average that disturbing given students are also reading for their classes and may have homework. The more disturbing version of this basic statistic is the variability. Nearly 20% of adolescents indicate they read nothing beyond what is assigned at school. Recent data on adult behavior indicates that the average daily reading time is about 15 minutes. Adults don’t set a very good example.

Summary

Reading both is an important source for what we know and what we know benefits the level of reading proficiency we achieve. Reading is an activity we control as individuals and if we so choose we can benefit from spending leisure time in this way.

References:

Cunningham, A. E., & Stanovich, K. E. (1997). Early reading acquisition and its relation to reading experience and ability 10 years later. Developmental psychology, 33(6), 934.

Recht, D. R., & Leslie, L. (1988). Effect of prior knowledge on good and poor readers’ memory of text. Journal of Educational Psychology, 80(1), 16–20. https://doi.org/10.1037/0022-0663.80.1.16 (baseball)

Rideout, V., Peebles, A., Mann, S., & Robb, M. B. (2022). Common Sense Census: Media use by tweens and teens, 2021. San Francisco, CA: Common Sense Media

Schneider, W., Korkel, J., & Weinert, F. E. (1989). Domain?specific knowledge and memory performance: A comparison of high? and low? aptitude Children. Journal of Educational Psychology, 81( 3), 306–312. (Soccer)

Willingham, D. T. (2017). The reading mind: A cognitive approach to understanding how the mind reads. John Wiley & Sons.

The Science of Reading: A Review of Three Recent Books

I read a lot, but aside from posting short reactions to Goodreads I seldom write longer reviews. I was able to get the book club I attend that typically focuses on history, economics, and writing to read my suggestion – Adrian Jones’ The Science of Reading. Despite the departure from our standard fare, the breadth of the issues covered using reading and learning to read as a base resulted in many engaging discussions. The present political interest of some states in mandating how children should be taught to read, the history of the “reading wars”, the big money up for grabs in school purchases of instructional materials (books), how digital technology changed reading, and why this particular skill is so important have encouraged broader interest in how best to prepare readers and in how we all engage with text. There is something for nearly everyone here.

The three books I will comment on cover these topics with different emphases. The Science of Reading (Jones) offers the strongest historical perspective. Reader, Come Home (Maryann Wolf) emphasizes a brain-based perspective. Willingham’s The Reading Mind approaches the set of topics from a cognitive perspective. All are approachable and as an educational psychologist with a cognitive background, I see value in how different perspectives support and supplement each other.

Adrian Johns’s “The Science of Reading: Information, Media & Mind in Modern America”

Johns covers the scientific study of reading from the 1880s to the present and argues that understanding the history of the science of reading is essential to understanding broader historical changes in knowledge, information, and technology. He achieves this by tracing the evolution of how reading has been perceived and taught, particularly in the United States, where reading has been closely linked to concerns about effective participation in a democracy. The assumed connection between reading and the economy and reading and informed citizenship are reflected in the long-term interest of politicians in this subject area.

The book begins by examining the origins of the science of reading, which can be traced back to the late 19th century, and the interest of scientists in the movement of the eyes (saccades) while reading. Early researchers like James McKeen Cattell (a name many might recognize from an Introduction to Psychology class) focused on the psychophysical aspects of reading, conducting experiments to measure reaction times and investigate how the human eye processes visual information. Cattell’s research led him to advocate for teaching reading through whole-word identification, a method that gained traction in American schools. For those involved in or recognizing the battle over how best to teach reading, the alternative extremes are to emphasize word recognition or sounding out words by what most would describe as phonics.

However, as Johns illustrates, the science of reading was not solely driven by laboratory findings. An emphasis on mass literacy further fueled the development and adoption of different reading methods, reflecting the intertwined nature of scientific inquiry and societal needs.

Edmund Burke Huey’s influential 1908 book, “The Psychology and Pedagogy of Reading”, marked a crucial shift in the field, arguing that reading was not merely a mechanical skill but a complex social activity shaped by readers’ experiences. Huey emphasized the importance of “apperceptive filling in,” where readers constantly make inferences based on clues from the text, highlighting the active and constructive nature of reading comprehension. Huey remained influential and my own introduction to reading as a fascinating area of study included my reading of an updated edition of Huey’s book in the 1960s. The word recognition position became integrated with the use of context to form one of the two alternative positions proposing how kids should be taught to read.

Johns dedicates considerable attention to the “reading wars,” the ongoing debates between proponents of phonics-based instruction and those favoring whole-language approaches. He argues that these debates, often framed as a simplistic this or that options, fail to capture the complexity of reading acquisition and the nuances of effective teaching methods. However, as one might recognize from recent media accounts of mandated methods in different states, the methods to be used and the commercial materials to be purchased with state money are often described to parents and the general public in similar simplistic ways.

The book examines key figures and events in the reading wars, including Rudolf Flesch’s scathing critique of whole-language instruction in his 1955 bestseller, “Why Johnny Can’t Read”. Flesch’s book ignited a public debate, pushing back against the prevailing emphasis on whole-word recognition and advocating for a return to phonics-based instruction.

Johns also discusses the work of Jeanne Chall, whose research, culminating in her 1967 book “Learning to Read: The Great Debate,” offered a more complex perspective on the reading wars. Chall argued that a balanced approach incorporating both phonics and whole-language strategies was crucial, particularly in the early stages of reading development. I was pleased to see that my friend Dick Anderson who once headed the Center for the Study of Reading at the University of Illinois also received some space in Jones’ book as a supporter of a similar balanced approach.

The latter part of the book explores the impact of technology on reading, examining the development of teaching machines, programmed learning, and the rise of computers. Johns discusses the work of visionaries like Alan Kay, whose Dynabook project, though never fully realized, envisioned a portable computer designed to replace school textbooks and transform the learning process. Today’s iPad could represent the type of device Kay could only imagine.

Johns concludes by reflecting on the enduring challenges of reading in the digital age, posing questions about the differences between reading on paper and screens, and the implications of new technologies for literacy and learning.

While providing a thorough historical overview, Johns occasionally delves into intricate details of specific experiments or research methodologies, which may not be of interest to all readers. I find the description of actual studies of great interest and the answer to the question of how researchers study something as invisible as the processes of a mental skill such as reading and come up with explanations of what exactly is going on.

Reader, come home: The reading brain in a digital world

Maryanne Wolf’s *Reader, Come Home* presents a critical examination of how the digital age is reshaping the human brain’s ability to read deeply. She reviews the brain science of her previous book (Proust and the Squid), but then spends time on whether new technologies presenting text and multimedia alter how the brain adapts to the processing of text. There is more brain science in her first book for those wanting that focus and more focus on screen time, changes in attention span, and reading from a screen versus from paper in the book I am reviewing here.

The central argument of *Reader, Come Home* is that the human brain was not initially designed for reading; instead, reading rewired our brains in ways that changed our thinking. Wolf is deeply concerned that excessive exposure to screen-based media is now rewiring our brains in a different, less beneficial way. The skimming style of reading encouraged by digital platforms, where information is processed quickly and often superficially, is contrasted with the deeper, more reflective reading associated with print books. Wolf stresses that while the brain is highly plastic and can adapt, the quality of that adaptation depends on the stimuli it is exposed to regularly.

Wolf does not argue for a wholesale rejection of digital reading. Instead, she advocates for a balance, calling for the development of what she terms “biliterate” brains. These are brains that are adept at both skimming digital media and engaging in deep, reflective reading. Her emphasis is particularly on children, who, she argues, are especially vulnerable to the negative impacts of digital reading due to the greater plasticity of their brains. If children are exposed only to skimming on digital platforms, they may miss out on the profound cognitive and linguistic benefits that come from deep reading.

The book also provides practical advice for parents and educators. One of Wolf’s key suggestions is to encourage parents to read to their children from printed books rather than relying on digital devices that often come with built-in distractions. The concern is not just about reading comprehension, but also about how children learn to engage with and reflect on complex information, skills that are crucial for higher-order thinking.

A recurring theme in Reader, Come Home is that we are becoming passive consumers of information rather than active, critical thinkers. Wolf’s fear is that if we continue down this path, future generations will lose the ability to think deeply and critically. She draws on cognitive science to show how the brain’s attentional systems are being rewired in ways that diminish our capacity for sustained attention, a vital component of deep reading.

For those who enjoy scientific debates, contrast Wolf and Willingham’s perspective on this issue. Willingham explains the resistance to reading long-form content more as a decreased tolerance for boredom rather than the brain being shaped biologically.

The Reading Mind: A Cognitive Approach to Understanding How the Mind Reads

The Reading Mind is a comprehensive exploration of the mechanics of reading, offering both a cognitive and a practical perspective on how our brains process text. Willingham draws from a wealth of psychological research to explain the nuances of reading, from identifying phonemes to building meaning from full texts.

At its core, The Reading Mind breaks down reading into stages, beginning with letter and sound recognition and culminating in full comprehension of texts. Willingham explains that the ability to match letters to phonemes (basic units of sound) is crucial to reading development, particularly for children. As readers become more proficient, their ability to decode words becomes more automatic, freeing cognitive resources for understanding the meaning of sentences and paragraphs. In other works, the sound of words is initially quite important, but gives way to more automatic recognition of words. This cognitive model serves as a foundation throughout the book as Willingham discusses the importance of both bottom-up processes, like letter recognition, and top-down processes, such as using background knowledge to comprehend texts. I read somewhere that this simultaneous bottom-up and top-down processing can be described as interactive, compensatory processing. This means that multiple processes are going on simultaneously and they work both to support each other. Letter recognition is information by word recognition (the word provides a context that speeds up letter recognition). Words are more quickly recognized and their meaning is more effectively retrieved within meaningful sentences (sentences provide a context that informs word recognition and understanding). Understanding what we read is assisted by what we already know in general and what we know more specifically about the topic we are reading about (again a context effect improving more basic processes).

One of the more powerful demonstrations I know of considers the relative contribution of reading proficiency and subject knowledge to comprehension. Willingham provides an example based on soccer knowledge. The research study I was familiar with used baseball knowledge. Anyway, young readers were classified as more and less proficient readers and more and less knowledgeable of baseball. This gives you four groups – high skill, high knowledge; high skill, low knowledge; low skill, high knowledge; and low skill, low knowledge. All readers were asked to read a description of half an inning of a baseball game and later write what they remembered. This method allows the impact of reading skill to be teased apart from the impact of subject knowledge. The findings demonstrate that existing knowledge had a larger impact on recall than reading skill.

One of the book’s strengths is its emphasis on the role of motivation in reading. Willingham makes the case that motivation is as critical as cognitive skills when it comes to becoming a proficient reader. Readers who enjoy reading tend to read more, which in turn improves their reading abilities, creating a virtuous cycle. This insight is particularly valuable for educators and parents trying to encourage reluctant readers. Willingham argues that ease of access makes a big difference

A notable section of The Reading Mind addresses the potential impact of technology on reading habits. Willingham acknowledges the concern that digital distractions, such as social media and video games, might reduce the amount of time young people spend reading. However, he points out that these activities have not necessarily displaced reading time for most youth. Instead, he suggests that the issue lies in a reduced tolerance for boredom, which could make sustained reading more difficult for some. While Willingham does recommend limiting screen time, he is skeptical of alarmist claims that technology is fundamentally altering how children’s brains process information. His balanced view on this topic is refreshing in an age of widespread concern about the effects of digital media on cognitive development.

Conclusion:

My efforts here were to give enough of the flavor of each of these books to perhaps convince you to take a look. As I tried to suggest in the beginning of this post, all books are very approachable and take you in multiple directions addressing several topics that seem of current interest.

ChatLLM from Abacus AI

I have this issue with AI services. I have identified several services that provide me useful services (image generation for posts, edits for what I write, identification of journal articles relevant to my writing, and chats with my large collection of notes and highlights) The issue is no individual tool does a great job of all of these things and I don’t use any of the several individual tools enough to justify the $20 a month that seems to be the going rate for any given tool. I could keep telling myself that this retirement writing gig should be treated as a hobby that I can afford, but there is a principle involved here. I think of many of my posts as exploring tools for teachers and $75-100 a month in their situation is likely a bit much.

I have found a general-purpose AI service, the ChatLLM tool from Abacus AI, that I have for me what is unlimited use for $10 a month. For a month or so now, I have been duplicating my use of ChatLLM and ChatGPT and can find very little difference in what the tools generate. Abacus.AI has an introductory offer that is a little confusing. The first month is free, but you must purchase one month and then you can cancel after the first two months.

The interface is very similar to what you experience with other AI services. There is a prompt field and responses appear above. ChatLLM allows the individual series of chat interactions with a given purpose allowing you to return if you have a reason to pick up again on an earlier exploration.

ChatLLM appears to function as an API allowing you to select from different AI services. This option seems useful for those of us who like to explore the possible differences among the AI services that are available.

AbacusAI offers many features I have not bothered to explain here in detail. I have learned what I thought necessary to meet my personal needs and it should be very easy for anyone with any experience with other platforms to get started. Here is a more complete review by another writer.