The Screen vs. Play Debate: What Science Says About Digital Toys, Educational Apps, and Whether They Actually Work
The app promises to teach your two-year-old to read. It has bright colours, cheerful music, and a rating of four and a half stars from seventeen thousand parents. It costs four euros per month. The reviews say things like “my daughter loves it” and “she asks for it every day” and occasionally “she learned so much.”
And you are standing in the app store, on your phone, at eleven o’clock at night, trying to decide whether this is something that will genuinely help your child or something that will sit on the tablet for a week, rarely used, while your child goes back to whatever they were doing before you introduced it.
The educational technology industry is one of the fastest-growing sectors in consumer technology, and it has built its expansion on the premise that screens can be educational — that the right app, the right program, the right interactive game can teach children things that books and physical toys cannot, faster and more engagingly. The market has responded to this premise with an enormous quantity of products, and parents have responded to the market with an enormous quantity of money and hope.
The research has a different story to tell. Not a simple one — not “all educational technology is ineffective” or “screens cannot teach children anything.” The actual research picture is considerably more nuanced than either the technology industry or its critics typically acknowledge, and understanding it changes what is worth downloading and what is not, and why.
Table of Contents
What “Educational” Actually Means
The first problem with the educational app category is definitional. Almost nothing is legally required before a technology product labels itself “educational.” The word is marketing, not certification. An app that flashes colours and plays sounds in response to a toddler pressing the screen is described as educational if the marketing team decides it is, regardless of whether any learning actually occurs.
The research definition of educational value is more demanding: does interaction with this material produce durable changes in knowledge, skill, or understanding that generalise beyond the specific interaction? A child who can press the correct button when a letter appears on screen may not have learned to read. A child who can score points in a mathematical game may not have developed mathematical understanding. The ability to perform within the specific context of the app does not necessarily indicate that generalised learning has occurred.
This distinction matters because most educational app research — including much of the research cited by developers — measures in-app performance rather than transferred learning. A study showing that children score higher on the app after using it for a month is not a study showing that they have learned anything that will be useful outside the app. Studies that measure genuine learning transfer — can the child apply what they apparently learned in the app to a different context, without the app’s scaffolding — are significantly rarer and significantly less encouraging.
The Video Deficit Effect
One of the most important and least widely known findings in the research on young children and screens is what developmental psychologists call the video deficit effect.
The video deficit effect describes the consistent finding that children under approximately thirty months learn significantly less from video presentations of information than from the same information presented by a live person. When a live person demonstrates how to perform a task — stacking blocks in a particular way, using a tool, finding a hidden object — toddlers can imitate and learn effectively. When the same demonstration is presented on a video screen, learning is significantly reduced, and in some studies is absent entirely.
The implication is significant: the intuitive assumption that a video of someone doing something has the same educational value as watching that person do it in real life is wrong, for children in their first two to three years of life. Something about the two-dimensional, non-contingent nature of the video presentation — the fact that it does not respond to the child, does not adjust to what the child does, does not provide the social and tactile cues that real interaction provides — reduces its effectiveness as a teaching medium for very young children.
The video deficit effect diminishes as children get older. By around three to four years, children show more effective learning from video, and by school age the gap between video and live presentation has largely closed for specific types of learning. But for the age group for which most “educational” apps and programmes are most aggressively marketed — infants and toddlers — the research indicates that live interaction is significantly more effective for learning than screen-based content, regardless of the quality of that content.
What Educational Apps Can and Cannot Do
The research on specific categories of educational technology is more differentiated than the general debate suggests, and it is worth being specific about what works and what does not.
What does not work well
Passive educational video — the programmes and apps that present educational content in a linear, non-interactive way — shows consistently poor learning outcomes in the under-three age group, for the reasons described above. The video deficit effect means that a child watching an educational programme about letters is learning significantly less than the same child being read to or exploring letter blocks with a caregiver.
Apps that produce the appearance of reading without the substance — that teach letter recognition and letter sounds through repetitive drill in a game format — sometimes produce children who can perform well within the app without having developed the phonemic awareness, vocabulary, and language comprehension that real reading requires. These apps may not be teaching reading. They may be teaching how to play the reading app.
Physical toy simulations on screens — virtual building, virtual drawing, virtual sorting — produce developmental benefits that are significantly reduced relative to the physical equivalent. The child who builds virtually is doing something cognitively interesting but is missing the fine motor development, the proprioceptive experience of physical materials, and the spatial understanding that comes from interacting with three-dimensional physical objects. The virtual version may look like the same activity. It is not developing the same things.
What works better
Video calling is the most consistently evidence-supported screen use for very young children, and the one most clearly distinct from other screen activities. Young children can learn language and maintain relationships with familiar people through video calling in ways that parallel face-to-face interaction, because the interaction is contingent — the person on the screen responds to what the child does and says. A child video calling a grandparent is having a qualitatively different experience from a child watching a video, because the grandparent responds.
Co-use with a caregiver — an adult who watches educational content alongside the child, points things out, asks questions, relates content to the child’s experience, and extends the interaction into real-world activities — dramatically improves the educational value of screen content. The video deficit effect is significantly reduced when a parent is actively engaged with the content alongside the child. The child who watches a programme about colours with a parent who then says “look, that flower is the same red as the one on the screen” is having a materially different educational experience from the child watching alone.
Interactive, contingent apps — those that genuinely adapt to what the child does, that respond to the specific child’s performance rather than playing a fixed sequence regardless of how the child responds — show better learning outcomes than passive content. The contingency — the experience of the app responding to what you do — partially compensates for the absence of human social response. This is what distinguishes genuinely interactive learning software from passive video dressed up as interactive.
Educational apps in middle childhood — roughly six to twelve years — have better evidence of effectiveness than those aimed at younger children, partly because the video deficit effect is largely gone and partly because the executive function and metacognitive capacities needed to learn deliberately from a digital medium are more developed. Older children can use well-designed educational software to genuinely learn things, in ways that toddlers with the same software do not.
| Category | Age Group | Evidence of Learning Effectiveness | What Makes It More Effective |
|---|---|---|---|
| Video calling with known person | All ages including infants | Good — contingent interaction transfers learning and relationship | Interactive, responsive, known person |
| Passive educational video | Under 3 years | Poor — video deficit effect; significantly less effective than live interaction | Co-viewing with active adult engagement improves it substantially |
| Interactive learning apps (genuinely contingent) | 3 years and above | Moderate — better than passive; still less effective than high-quality physical alternatives | Genuine contingency; transfer testing; adult involvement |
| Letter/number drill apps | Under 5 years | Limited — in-app performance does not reliably transfer to real reading/maths | Limited improvement regardless of engagement level |
| Educational software for school-age children | 6 to 12 years | Moderate to good — transfers better; executive function available to use it | Well-designed adaptive software; connected to curriculum; not replacing physical learning |
| Virtual physical activity simulation | All ages | Significantly reduced relative to physical equivalent | Cannot substitute for the developmental value of physical material interaction |
Why Children Engage With Apps That Don’t Teach Much
One of the most counterintuitive things about the educational app research is the gap between engagement and learning. Children who appear very engaged with an educational app — who ask for it repeatedly, who spend extended time on it, who seem absorbed — are not necessarily learning much from it. High engagement is not the same as effective learning.
The apps that produce the highest engagement tend to share certain features: fast feedback, bright visual effects, sound rewards, variable reward schedules (you don’t know exactly when you’ll get the reward, which is the most addictive reinforcement schedule known to psychology). These features are engineered to maximise time on app — because time on app is the metric that matters for the app’s commercial success. They are not engineered to maximise learning, which often requires slower, more effortful processing that is less immediately rewarding.
This means that the child who loves the maths app and asks for it every day may be doing something closer to playing a well-designed game than to learning mathematics. The app has successfully produced the behaviours — engagement, return visits, enthusiasm — that look like learning from the outside, without necessarily producing the cognitive change that learning actually is.
This is not universal. Some well-designed educational software produces both high engagement and genuine learning. But the correlation between how much a child wants to use an app and how much they are learning from it is much weaker than the app’s marketing would suggest.
The Physical vs. Digital Comparison
The most practically useful question for most parents is not “does this app teach anything” but “does this app teach more than the physical alternative would?” Because the opportunity cost of digital play is usually physical play, and the question of relative value matters.
The research on physical toys and play — particularly open-ended construction, art, outdoor play, and social pretend play — consistently finds developmental benefits that are broader, more durable, and better-evidenced than comparable digital activities. The child who spends thirty minutes building with blocks is developing spatial reasoning, fine motor skills, executive function, and creative thinking in ways that are measurable and well-supported by decades of research. The child who spends thirty minutes with a building-themed educational app is doing something that looks similar but is developing significantly fewer of these capacities, less completely.
This does not mean digital and physical activities cannot coexist. A child who has rich physical play experiences and some carefully chosen digital experiences is in a different position from one whose play is primarily or exclusively digital. The concern is not any specific use of educational technology. It is the systematic substitution of physical play with digital activity that the research suggests is developmentally costly.
The physical play that is most completely unreplaceable by digital activity is the play that is most sensory, most three-dimensional, and most socially interactive. Outdoor physical play, art and making, construction with physical materials, and social pretend play with other children — these are the activities that the developmental evidence is strongest for, and the ones for which digital alternatives are least adequate.
What Actually Makes Educational Technology Worth Using
Given the research picture, what distinguishes the digital activities that are worth including in a child’s life from those that are not?
Adult co-engagement is consistently the most powerful modifier of educational technology’s effectiveness. The same content, used by a child alone versus used with an engaged adult who extends and discusses what is happening, produces substantially different learning outcomes. The parent who sits with the child and turns a digital activity into a conversation — asking questions, drawing connections to real experience, extending the interaction beyond the screen — is making a mediocre educational app significantly better. The parent who uses the app as a substitute for engagement is making a potentially good educational experience significantly worse.
Contingency — genuine responsiveness to what the child does — distinguishes apps that function more like live interaction from those that are effectively passive video with a touch interface. The child whose input actually changes what happens next is in a different cognitive position from the one whose “interaction” is pressing buttons at predetermined moments in a fixed sequence.
Transfer testing is the standard that separates genuinely educational technology from technology that only appears educational. Can the child do something outside the app that they could not do before? Can they apply what the app apparently taught in a different context, with different materials? If the only evidence of learning is in-app performance, the learning may not have happened in the sense that matters.
Age-appropriateness in the neurodevelopmental sense, not the marketing sense. The age ranges printed on app packages reflect the complexity of the content, not the developmental readiness of the child to learn from that medium. A toddler is genuinely developmentally unsuited to learning effectively from passive screen content, regardless of how simple and age-appropriate the content is. Understanding the video deficit effect — and its developmental timeline — provides a better guide to which age groups can genuinely benefit from educational technology than any marketing claim.
For the broader research picture on screens and child development — including the sleep effects, the social media dimension in adolescence, and the overall framework for calibrating screen use to developmental priorities — the article on what the research actually says about screen time provides the complementary context that this article’s specific educational technology focus does not cover. And for the specific research on open-ended physical play and why simple materials consistently outperform sophisticated educational toys in developmental terms, the article on the building block principle makes the case for what digital technology is competing against.
A Practical Framework for Evaluating Educational Apps
Rather than a list of recommended apps — which would be outdated before this article was published — here is a set of questions that apply to any educational technology product and that the research supports as meaningful discriminators between effective and ineffective educational tools.
Does it require the child to actively generate responses, or does it primarily present content for the child to receive? Active generation — answering questions, solving problems, creating something — is more likely to produce learning than passive reception.
Does it respond differently to different children’s inputs, or does it play a fixed sequence regardless of what the child does? Genuine contingency indicates more sophisticated design and produces better learning.
Does it have any evidence of learning transfer — studies showing that children who used it performed better on measures outside the app than those who did not? Be sceptical of evidence that only shows in-app performance.
Does it connect to something real — a book, a physical activity, a real-world experience that the digital content relates to? Technology that is an extension of a rich physical and social environment is different from technology that substitutes for it.
Could the same learning goal be achieved as well or better with a physical alternative? If yes, the physical alternative is probably worth choosing, particularly for younger children, for the developmental breadth it provides.
| Question | Why It Matters | Good Sign | Concern |
|---|---|---|---|
| Does the child actively generate, or passively receive? | Active generation produces more durable learning | Child must answer, create, or solve to progress | Child primarily watches and responds at predetermined moments |
| Is there genuine contingency? | Response to child’s specific input distinguishes interactive from passive | Difficulty adapts; different paths for different responses | Same sequence regardless of child’s input |
| Is there evidence of learning transfer? | In-app performance is not real learning if it doesn’t transfer | Independent studies showing transfer to other contexts | Only evidence is in-app scores or parent testimonials |
| Does it connect to the physical world? | Technology embedded in rich environment is more effective than technology replacing it | App extends or relates to physical activities, books, real experiences | App is self-contained with no connection to anything outside it |
| Is the engagement rewarding learning or rewarding time-on-app? | Engagement features can be designed for addiction rather than learning | Rewards reflect genuine achievement; difficulty increases meaningfully | Constant sound effects and visual rewards regardless of learning |
What This Means for Your Family
The research picture is not “delete all apps and return to wooden toys.” It is considerably more nuanced, and it points toward specific principles rather than categorical rules.
Under three years: the developmental case for educational apps is weak, and the case for physical play, face-to-face interaction, and outdoor exploration is strong. If screens are used, video calling with known people and co-viewing of high-quality content with an engaged adult are the best-supported uses. The apps that promise to teach your eighteen-month-old to read are not delivering what they promise.
Three to six years: some educational apps have genuine value, particularly those with strong contingency, connection to physical activities, and adult involvement. The best educational technology at this age supplements a rich physical play environment rather than substituting for it. The question to ask is always: what is this replacing, and is the replacement worth it?
Six years and above: the evidence for educational technology is better, and well-designed software can support learning in ways that are genuinely effective. The priority at this age shifts from the medium itself to the quality and purpose of the content: is the child using technology to learn something meaningful, or to be entertained in a way that is being marketed as educational?
At all ages: adult co-engagement is the most powerful lever available for increasing the developmental value of any screen-based activity. Twenty minutes with a parent who is genuinely engaged with what is on the screen is worth more than two hours alone.
Frequently Asked Questions
Are there any educational apps that actually work for very young children?
The research suggests that very few apps produce genuine learning transfer in children under three, because of the video deficit effect — the consistent finding that toddlers learn significantly less from screen presentations than from live interaction. The exception is video calling with known people, which appears to produce genuine social and language learning. For children over three, some well-designed adaptive software shows genuine learning benefits, particularly when used with adult co-engagement. The key question is always whether the child can do something outside the app that they could not do before.
My child learned their letters from an app. Doesn’t that prove it works?
Possibly, though the causation is hard to isolate — children who use educational apps also typically have parents who engage with their learning in other ways. More importantly, letter recognition is one component of reading, and learning to recognise letters in an app context does not necessarily produce the phonemic awareness, vocabulary, and comprehension that actual reading requires. Whether the app taught letter recognition or whether the child would have learned it from other engagement is difficult to know. Whether the app taught reading is a different question, and one for which the evidence is less encouraging.
My child loves their educational app. Doesn’t that mean it’s good for them?
High engagement is not the same as effective learning. Apps are engineered for engagement, and the features that produce engagement — fast feedback, variable rewards, bright visual effects — are not the same as the features that produce learning. A child can be highly engaged with an app that is teaching very little, in the same way they can be highly engaged with a television programme that is entertaining without being educational. Asking what the child can do outside the app that they could not do before is a better measure than how much they want to use it.
How do I find educational apps that actually have research evidence behind them?
Look for apps that have been independently evaluated — not by the developer, but by researchers with no financial interest in the outcome. University-based technology and early childhood development centres publish evaluations of specific products. The Joan Ganz Cooney Center, associated with the Sesame Workshop, produces evidence-based reviews of educational media for children. Common Sense Media rates apps on multiple dimensions including educational value. Be sceptical of research that only appears on the developer’s own website and that only measures in-app performance rather than transfer to other contexts.
Is it better to have no apps at all for young children?
The research does not support a zero-tolerance position on technology for young children. What it supports is prioritising the activities with the strongest developmental evidence — physical play, face-to-face interaction, outdoor exploration, reading together — and treating screen-based activities as supplementary rather than central, particularly under three years. The child who has rich physical and social play experiences and some carefully chosen screen time is in a good position. The concern is the systematic replacement of physical and social play with screen time, not the presence of any screen time at all.
What about coding apps and STEM apps for older children?
These have better evidence of effectiveness than most educational technology aimed at younger children, for several reasons: they typically require active problem-solving rather than passive reception, they have genuine transfer potential to programming and computational thinking, and they are being used by children whose cognitive development is better suited to learning from digital media. The strongest outcomes are in children who use these tools in a context that connects to physical making and real-world application — coding that controls a robot, programming that produces something visible — rather than purely abstract digital environment.
Younes Kehal is a Professional Educational Director and School Coach with over 20 years of experience working directly with children, families, and educational institutions. The guidance published on Parenting Assist is rooted in real field experience and evidence-based developmental science.
