Researchers around the world have been testing the ideas behind First Cog for years: hands-on foundations first, screens later, and teaching kids how machines learn before they lean on one. Here is what they found. Every link goes straight to the source.
Dr. Jared Cooney Horvath, a neuroscientist and educator who studies how people learn, testified before the Senate Committee on Commerce, Science, and Transportation at its hearing Plugged Out: Examining the Impact of Technology on America's Youth. He told the committee that more screen exposure in classrooms is generally associated with weaker learning outcomes, not stronger ones.
"It is a question of aligning educational tools with how human learning actually works."
That's the design brief behind First Cog. Human learning first, tools second.
Watch Dr. Horvath's testimony · Open on YouTube →
Jared Cooney Horvath · Harmony / Penguin Random House
Horvath's book-length case on classroom technology, which his publisher describes as "a practical playbook for putting people, not programs, back at the center of education."
Every First Cog session is people at the center: a teacher, a group, a table and a problem worth arguing about.
Horvath, Horton, Lodge & Hattie · First Monday 22(9) · Free to read
Content taken in all at once felt well understood at first, then faded faster than the same content spaced out daily or weekly.
First Cog runs as a sequence, not a cram. Each week builds on the last, so ideas are revisited until they stick.
Horvath & Donoghue · Frontiers in Psychology · Free to read
Makes the case that teaching should be built on the science of how people actually learn, carefully translated into classroom practice.
First Cog's sequence follows that science: concrete, hands-on experience first, abstraction and tools after.
Meta-analyses, controlled experiments and classroom studies from around the world point the same way: kids who learn computer science with their hands build stronger thinking, more confidence and more motivation.
Chen, Yang, Metwally, Lavonen & Wang · International Journal of STEM Education 10, 47 · Free to read
Reviewing 49 K–12 studies, the authors found a large overall effect of unplugged activities on computational thinking (Hedges' g = 1.028), with strong results in primary school. Board and card games were among the most common tools.
Cards, boards, strings and beads are the working materials of every First Cog session.
Zhang, Liang, Tian & Yu · Educational Technology Research and Development
Fifteen experimental and quasi-experimental studies, combined: unplugged programming activities have a positive effect on K–9 students' computational thinking (g = .631). The authors call for more storytelling and collaborative unplugged activities in the classroom.
Collaboration and story run through First Cog. Kids work as a team to become the network, the algorithm and the message.
Hermans & Aivaloglou · WiPSCE 2017, TU Delft
Children who started unplugged mastered the same programming concepts as children who started on screens, and came away with better self-efficacy, using a broader range of programming blocks once they reached Scratch.
Confidence facing a problem they've never seen is the outcome First Cog leads with. Unplugged first is how it gets there.
Bell & Vahrenhold · Adventures Between Lower Bounds and Higher Altitudes, Springer LNCS 11011
The co-creator of CS Unplugged surveys how the approach is used worldwide and what the research shows, including stronger self-efficacy when unplugged lessons come before programming.
First Cog is built on that order: understand the idea with your hands, then build with the tools.
Tim Bell · Communications of the ACM · Free to read
Bell describes unplugged learning as "a catalyst that supports learning to program," one that "can be used to improve self-efficacy in students without increasing teaching time or decreasing programming skills."
That is exactly how First Cog uses it: the catalyst for everything that follows, coding and AI included.
del Olmo-Muñoz, Cózar-Gutiérrez & González-Calero · Computers & Education 150
Second graders who began with unplugged activities before moving to computers improved more in computational thinking, and were more motivated, than classmates who worked only on screens.
Unplugged first, screens later. It's the First Cog sequence, studied in a real classroom.
Brackmann, Román-González, Robles, Moreno-León, Casali & Barone · WiPSCE '17, ACM
Students ages 10 to 12 who took unplugged lessons made significant gains on a validated Computational Thinking Test.
Ages 10 and up are the heart of the core First Cog program.
Education and Information Technologies
Sixth graders taught with unplugged instruction reached higher basic programming achievement than peers taught on computers.
Understand the idea first and the code comes easier. First Cog builds that understanding before anyone touches a keyboard.
Li, Jiang, Liang, Xiong, Liang & Zhao · Education and Information Technologies 28(9)
Students in interactive, hands-on unplugged lessons performed considerably better in computational thinking and engagement than students in non-interactive lessons.
First Cog sessions are built around participation. Nobody watches from the side.
Delal & Oner · Informatics in Education 19(1) · Free to read
Sixth graders working through unplugged problem-solving tasks from the international Bebras challenge significantly improved their computational thinking, with girls and boys gaining alike.
A puzzle they've never seen is how every First Cog session begins.
Thinking Skills and Creativity
A large study of children ages 5 to 9 found unplugged methods both effective and practical, with trained teachers designing activities for their own classrooms.
First Cog's 45-minute version for ages 7–9, and its teacher training, put this into practice.
Battal, Afacan Adanır & Gülbahar · Journal of Educational Technology Systems 50(1)
Fifty-five studies show the range of the unplugged approach: paper-and-pencil activities, problem solving, storytelling, games and tangible materials.
First Cog draws on all of them, chosen to fit the group in the room.
Huang & Looi · Computer Science Education 31(1)
Finds that unplugged pedagogies have opened computer science to K–12 students who might otherwise never meet it, and could do the same inside formal schooling.
An entry point anyone can reach and a ceiling that's hard to find. That's First Cog.
Hu, Huang & Li · Early Childhood Education Journal 52(4)
Children's computational thinking "extended to both the unplugged toolkit and children's bodies," emerging as they moved, handled and acted out ideas.
In Build a Machine That Learns, kids don't watch a neural network. They become one.
Lee, Martin, Denner, Coulter, Allan, Erickson, Malyn-Smith & Werner · ACM Inroads 2(1)
Introduced the Use-Modify-Create progression: learners use an idea, then change it, then create their own.
Every First Cog module climbs the same ladder.
Sentance, Waite & Kallia · Computer Science Education 29(2–3)
Students who learned to predict and investigate before writing their own programs outperformed a control group, and their teachers grew more confident teaching programming.
Understand before you build. It's the order First Cog keeps from the first session to the last.
Busuttil & Formosa · Informatics in Education · Free to read
In secondary-school classrooms, unplugged computing proved effective at engaging students, drawing them into the work, and building teamwork and collaboration.
First Cog accelerates for older learners, and every challenge is a shared problem the whole group takes on together.
Dağ et al. · Journal of Computer Assisted Learning
An unplugged course for 212 primary school students improved five core computational thinking skills: algorithmic design, abstraction, evaluation, decomposition and generalization.
Breaking problems down, spotting patterns, testing ideas: First Cog modules put all five to work, week after week.
Looi, How, Wu, Seow & Liu · Computer Science Education
Follows students from a hands-on sorting activity into written descriptions, flowcharts and code, tracing how the physical activity builds computational thinking.
First Cog gives every idea a concrete form first, so when a group reaches code, the code already means something.
Education and Information Technologies
Shows unplugged activities and block-based programming working together to strengthen computational thinking.
Understand → Build → Use AI. First Cog's path puts unplugged foundations exactly where they do the most good.
Some of the most-cited research in learning science, from active learning to spaced practice, points straight at the way First Cog teaches.
Freeman, Eddy, McDonough, Smith, Okoroafor, Jordt & Wenderoth · Proceedings of the National Academy of Sciences 111(23) · Free to read
A meta-analysis of 225 studies of undergraduate science, engineering and math courses found that students in active-learning classes scored higher on exams and were less likely to fail than students in traditional lectures.
First Cog puts learners to work on the thinking itself: ask, try, explain, test, improve. Participation isn't an add-on. It is the lesson.
Deslauriers, McCarty, Miller, Callaghan & Kestin · Proceedings of the National Academy of Sciences 116(39) · Free to read
In university physics classrooms, students learned more through active engagement than through polished lectures, even though they felt they had learned less.
Real learning can feel like effort. First Cog challenges ask kids to decide, defend and rethink, and that effort is where the understanding comes from.
Chi & Wylie · Educational Psychologist 49(4)
The ICAP framework ranks engagement from passive to active to constructive to interactive, with interactive work, where learners build on each other's ideas, producing the deepest learning.
First Cog groups compare approaches, argue about results and build on each other's ideas. Interactive by design.
Dunlosky, Rawson, Marsh, Nathan & Willingham · Psychological Science in the Public Interest 14(1)
A landmark review of ten popular learning techniques rated practice testing and distributed practice as the most effective, across ages and abilities.
First Cog's progression keeps earlier ideas in play. Each module calls on what came before, so understanding gets used, not just covered.
Agarwal, Nunes & Blunt · Educational Psychology Review 33
Across 50 classroom experiments with 5,374 learners, recalling what had been learned improved learning across subjects and education levels.
To earn a First Cog badge, a kid teaches the idea back to someone else. That's retrieval, built into the reward.
Cepeda, Pashler, Vul, Wixted & Rohrer · Psychological Bulletin 132(3)
A synthesis of 317 experiments found that spreading learning out over time strengthens retention compared with packing it into one sitting.
Weekly First Cog sessions that build on each other give every idea time to take hold.
The largest education studies on the planet, from the OECD and UNESCO, along with two meta-analyses on paper versus screens.
OECD, PISA · Free to read
"PISA results show no appreciable improvements in student achievement in reading, mathematics or science in the countries that had invested heavily in ICT for education." Making sure every child masters the basics "will do more to create equal opportunities in a digital world than … expanding or subsidising access to high-tech devices."
First Cog invests in the thinking first. Devices come in when they're the right tool.
OECD · Free to read
Across OECD countries, 65% of students said they were distracted by using digital devices in at least some math lessons. Students distracted by classmates' devices scored 15 points lower in math.
There are zero devices in a First Cog session. All the attention goes to the problem on the table.
UNESCO · Free to read
UNESCO urges that technology "never replaces in-person, teacher-led instruction." In Peru, a program that handed out more than a million laptops produced no positive impact on learning, which the report links to a focus on devices instead of teaching.
First Cog is in person and teacher-led, every single time.
Delgado, Vargas, Ackerman & Salmerón · Educational Research Review 25
Pooling 54 studies and more than 171,000 participants, the authors found comprehension is better on paper than on screens for informational text, and the advantage has grown in recent years.
Paper, cards and pencils carry the work in First Cog.
Virginia Clinton · Journal of Research in Reading 42(2)
Readers performed better on paper than on screens, and judged their own understanding more accurately on paper.
Hands-on work makes understanding visible, to the learner and to the teacher. That's the point of every First Cog activity.
The field agrees that every child should know how AI actually works. The best way to teach it starts without a computer.
Touretzky, Gardner-McCune, Martin & Seehorn · Proceedings of the AAAI Conference on AI · Free to read
The founding paper of the national AI4K12 initiative, setting out the "Five Big Ideas in AI" every child should understand, including how machines learn.
Learning machines is the capstone layer of the First Cog Curriculum, and the tip students see first in the intro session.
Long & Magerko · CHI 2020, ACM
Defines AI literacy as the ability to "critically evaluate AI technologies; communicate and collaborate effectively with AI; and use AI as a tool," and recommends embodied, hands-on ways of learning it.
Question the answer. It's First Cog's headline promise, and the first competency on this list.
Yun Dai · Education and Information Technologies 30(5) · Free to read
Over two years with 107 sixth graders, embodied, unplugged analogies followed by hands-on projects built students' AI knowledge, skills and ethical reflection.
Understand it, build it, then use AI well. The same arc First Cog follows.
Lindner, Seegerer & Romeike · ISSEP 2019, Springer LNCS 11913
The "AI Unplugged" series teaches machine learning, classification and neural networks without a single computer.
Strings, beads and a tic-tac-toe board. First Cog's own take on the same idea.
Ma, Sanusi, Mahipal, Gonzales & Martin · SIGCSE 2023, ACM
Middle schoolers said unplugged decision-tree and nearest-neighbor activities helped them understand how machine learning algorithms work.
Kids who can explain how a machine decides are kids who can question what it decides.
The best time to start was before AI. The second best time is now.