My scientific career began in “natural scene analysis”. My entrepreneurial career began with naturalistic assessment using machine learning to turn student learning experiences directly into student assessments…no exams needed. My philanthropic work in health and human development has all been about understanding people in the natural pace of their lives and making changes that were just as natural. “Naturalistics” is the defining feature of my work.

Why?

Naturalistic methods capture behavior in real-world contexts, reflecting the complexities and nuances of everyday life. While people often alter their behavior when they know they are being observed (the Hawthorne effect) or when responding to explicit questions (social desirability bias), naturalistic methods, especially when surreptitious, minimize this reactivity.

Further, naturalistic settings allow for the observation of spontaneous, unprompted behaviors that might never emerge in a controlled environment or appear in a self-report survey. This is crucial for understanding the full range of heterogeneous human actions and reactions, including rare events, subtle interactions, and unconscious habits.

In fact, I happily take the inevitably messy data from most naturalistic methods (e.g., wearable sensors, digital trace data) over clinical-quality data because they provide a dynamic, longitudinal view of behavior, revealing patterns, changes, and individual differences that would be missed by snapshot assessments.

Everything above goes double for interventions…unfortunately, it’s also doubly hard. Understanding behavior in its natural context allows for the design of interventions that are tailored to the specific situations and challenges people face in their daily lives. Naturalistic data streams also enable "just-in-time" or even predictive interventions, which increases the likelihood of behavior change by providing support at moments of vulnerability or opportunity.

As naturalistic data reveals heterogeneity in behavior patterns and preferences, it also allows for the creation of personalized interventions that are more effective than fits-to-model approaches. One of the most powerful if underappreciated effects of this are interventions that are integrated into people's daily lives and routines. This seamless blend of interventions into the natural flow of behavior improves adherence and long-term outcomes.

The kinds of naturalistic methods I use—passive sensing or analysis of metadata—are less burdensome for participants than traditional assessments, which is a huge gain of intrusive workplace spark surveys. And because the data already exists, it also tends to be quite cost effective.

Naturalistic methods raise serious privacy concerns. It is always important to me to be completely transparent and opt-in. It is my job to make sense of the data, not the participants' job to fit their lives into my goals.

I’m willing to bear those ethical burdens, however, because the opportunity to holistically integrate biological, behavioral, and contextual data is the path to exploring the complexity of influences and interactions among people and tease apart correlation and causation.

Get messy. Get natural.

Follow me on LinkedIn or join my growing Bluesky!

Research Roundup

Brain Training Isn’t Meta-Learning

“There’s no trick to it; it’s just a simple trick!” Every company building gamified brain training programs is Brad Goodman.

Take “executive control”, for example: measured in childhood it predicts many positive life outcomes, and it would be great if there were a simple trick to developing it.  Unfortunately, a “highly motivating gamified interface” had “no impact on any behavioral outcomes…or neural outcomes”.

Despite “long-lasting improvements of closely related measures of cognitive control [i.e. not pressing a button] at the 1-year follow-up”, kids showed no improvement in decision-making, academic achievement, mental health, fluid reasoning, creativity, task-dependent and intrinsic brain function, nor gray and white matter structure.

Whether for education or health aging, “brain training” must be naturalistic and hard. Instead of simple tricks, go deep with meaningful thought: read a challenging book, learn a new language, explore the world.

Naturalistic Measures

When trying to understand psychology and marketing, there are lies, damn lies, and questionnaires. We can do better.

Students struggled during the pandemic, yet traditional survey measures of GRIT (a construct of resilience and perseverance) showed no “predictive power over performance changes” during lockdown. How could the very thing that predicts perseverance show no effect?

A novel “behavioral measure of GRIT” derived from actual student behavior prior to the pandemic revealed that “students who were grittier…before the pandemic, register lower declines in math and science scores during the coronavirus period”. Unlike the survey, this measure “can explain 77% of the variance in academic resilience”.

If we want to understand human lives we need naturalistic measures of their real behavior.

If we want to change human lives, we need naturalistic interventions to change their behavior.

"Metacognitive Laziness"

Hell Yes: “AI technologies such as ChatGPT may promote learners' dependence on technology and potentially trigger ‘Metacognitive Laziness’.”

Far too many believe that their new AI product will somehow change the world. Start by making certain it can even change 1 person. Students with access to ChatGPT (vs human experts or automated writing tools) asked the LLM more questions and wrote better essays…they didn’t learn any more than anyone else. Nor did they show any “difference in post-task intrinsic motivation”.

Given related research, most students using GPT become reliant on it, both internalizing its capabilities as their own and decreasing their self-assessment.

These findings are entirely in keeping with my predictions. The students with meta-learning strengths showed positive complementarity with AI tutors and those low in meta-learning showed negative complementarity, performing worse than the human expert or even control groups.

<<Support my work: book a keynote or briefing!>>

Want to support my work but don't need a keynote from a mad scientist? Become a paid subscriber to this newsletter and recommend to friends!

SciFi, Fantasy, & Me

I’ve been watching Severance with my son. It reminds me of watching the first few seasons of Lost with grad school classmates or the first season of Twin Peaks. Speculating with friends about the mystery box is just fun. Speculating while occasionally saying, “Whow, wait. Is the evil dentist…Robbie Benson?” So fun.

Now to stick the landing, Severance!

Stage & Screen

• March 27, Lawrence Berkeley Labs: Scientific innovation and the value of thinking differently.
• So many upcoming podcasts...and something even bigger!
• May 7, Chicago: Innovation, Collective Intelligence, and the Information-Exploration Paradox
• May 8, Porto: Talking about entrepreneurship at the SIM conference in Portugal
• May 14, London: it time for my semi-annual lecture at UCL.
• June 12, SF: Golden Angels
• June 9, Philadelphia: "How to Robot-Proof Your Kids" with Big Brothers, Big Sisters!
• June 18, Cannes: Cannes Lyons
• Late June, South Africa: Finally I can return. Are you in SA? Book me!
• October, UK: More med school education

If your company, university, or conference just happen to be in one of the above locations and want the "best keynote I've ever heard" (shockingly spoken by multiple audiences last year)?

Vivienne L'Ecuyer Ming