Six-month-old babies, who cannot yet say a single word, already seem to size up the people around them and prefer the kind ones. In a now-classic 2007 study published in Nature, Yale psychologists Kiley Hamlin, Karen Wynn, and Paul Bloom showed infants a simple wooden-block puppet show and found that a striking majority reached for the character who had helped another character, not the one who had blocked it.¹

The finding has been repeated, refined, and pushed back on for nearly two decades. What is left after all that picking and prodding is a careful but real claim: long before babies can talk, they are already paying close attention to who is nice, who is mean, and which side they want to be on.³

How do you ask a baby a moral question?

You cannot interview a six-month-old. So Hamlin and her colleagues built a stage. A round wooden Climber, with painted-on eyes, struggled to make it up a small hill. Sometimes a Helper character, a yellow square, came in from below and gently nudged the Climber to the top. Other times a Hinderer, a blue triangle, shoved the Climber back down.¹

After the babies watched the show several times, the experimenter held the two characters out, one in each hand, and waited. Reaching is the first verb a baby really has. It is how a baby says, “that one, please, the one I want closer.” In the original Nature paper, almost every 6-month-old and every 10-month-old who participated reached for the Helper.¹

Schematic illustration of a tiny puppet show stage seen at infant eye level: a wooden hill in the center, a round red Climber shape part way up the slope, a yellow square Helper figure pushing it up from below, and a blue triangle Hinderer figure pressing down from above. Glowing teal arrows trace the directions of force, and a faint outline of an infant's face watches from the side of the frame

That is a startling result for a population that mostly cannot sit up unsupported. The babies were not being told the Helper was good. Nobody clapped or cheered. The puppets did not have voices, costumes, or names. The only difference between the two shapes was what each had done in the seconds before. Even so, the babies sorted them, and they sorted them in the same direction the rest of us would.

What if babies just like things that go up?

Skeptics raised a fair point right away. Maybe babies were not making a moral judgment at all. Maybe they just preferred upward motion, or yellow squares, or whatever low-level visual feature happened to track with the helping role.

Hamlin’s group ran control conditions to chase that worry down. They removed the Climber’s eyes, so it looked like a piece of luggage rather than an agent with a goal. With no goal in the scene, the babies stopped showing a preference. The Helper-versus-Hinderer effect did not survive once the Climber was no longer perceived as someone trying to do something.¹ That argues the babies were tracking goals, not shapes.

A 2011 follow-up in Cognitive Development tested even younger participants and pushed the effect down to 3-month-olds, who looked longer at the prosocial puppet rather than reaching, since reaching is barely available at that age.³ A separate 2010 paper, also in the Hamlin program, reported that three-month-olds tend to show what the authors called a negativity bias: they were quicker to avoid the unhelpful character than to choose the helpful one.² Reading those two findings together, the gist is that very young infants notice both sides of the social ledger, and the avoid-the-jerk circuitry seems to come online a hair before the approach-the-friend circuitry.

Candid phone-snapshot of a Caucasian mother in her early thirties with shoulder-length light brown hair, wearing a cream knit sweater, sitting cross-legged on a beige rug and holding a six-month-old baby on her lap. The baby has fair skin and wispy blonde hair and is reaching toward a small wooden toy held by an unseen second adult. Natural afternoon window light, slightly grainy

Do babies actually want the bad guy to be punished?

This is where it gets uncomfortable. In a 2011 paper in the Proceedings of the National Academy of Sciences, Hamlin and her co-authors went a step further. After watching a Hinderer mistreat a third party, slightly older infants and toddlers preferred a new character who took a treat away from the Hinderer over a character who gave the Hinderer a treat.⁴

Read carefully: the babies were not just rooting for kindness in the abstract. They were rooting for the antisocial character to lose. By around 21 months, toddlers in the study even chose to physically remove a treat from the wrongdoer themselves when given the option.⁴ The authors are cautious about calling this “punishment.” Toddlers do not have a courtroom in their heads. But the behavior pattern is consistent with a very early version of the moral instinct most adults recognize: bad actors should not be rewarded.

Is it the action, or what the puppet was trying to do?

One of the more interesting wrinkles came in 2013, when Hamlin published in Cognition on what she called “failed attempts to help and harm.”⁵ If a Helper tried hard to nudge the Climber up but slipped and failed, did the babies still count it as a Helper? If a Hinderer lunged for the Climber to push it down but missed, did the babies still hold it against them?

The short answer is yes, with a developmental caveat. Eight-month-olds tracked intention, not just outcome. They preferred the would-be Helper even when the help did not land, and they avoided the would-be Hinderer even when the harm did not land.⁵ Five-month-olds, in contrast, were less consistent and seemed to lean more on what actually happened on screen. So the capacity to read intent appears to mature across the first year. By the time a baby is approaching their first birthday, they are not just running a simple “helped versus did not help” tally. They are running something closer to “tried to help versus tried to harm,” which is a much more sophisticated piece of social bookkeeping.

Cross-section diagram of an infant brain with the temporal-parietal junction and medial prefrontal cortex gently highlighted in glowing teal, soft amber neuron sparks tracing connections between them, the rest of the brain rendered in a low-contrast charcoal silhouette

Why would evolution build this in so early?

One reading of the data is that humans cannot afford to wait until language to start sorting their social world. A newborn is helpless for years. Whether the people around the baby are reliably going to feed, lift, soothe, and protect them is a life-or-death matter, and it would be strange if the brain did not bring some pre-installed equipment to that problem.

Hamlin and Bloom have argued that an early, automatic preference for cooperative behavior is exactly the kind of bias natural selection might favor. Babies who orient toward helpers, and away from hinderers, are more likely to end up in the good graces of cooperative caregivers and group members. The technical word researchers sometimes use is a “core” social-evaluation system: a relatively simple mechanism that gets the basics right, that is then refined by experience, language, and culture across childhood.³

That does not mean the system is fixed or fully formed. It is more like a rough first draft. By age two, toddlers begin to consider context, intentions, and the identity of the victim, all of which a 6-month-old mostly does not. The first-draft theory fits the data: there is a basic preference present very early, and richer judgments are layered on top as the years go by.

Has this finding been controversial?

Yes, and any honest summary has to say so. Several labs across the 2010s tried to replicate the original Hamlin Helper/Hinderer effect with mixed results, and a long-running scientific conversation has centered on which features of the puppet show babies are actually keying on, and how robust the preference is across cultures, materials, and labs.

Hamlin and others have responded with refinements, including a 2014 commentary in Frontiers in Psychology arguing that the babies’ preferences depend specifically on the Climber’s perceived goal, indexed by where it gazes. When the Climber clearly wants to go up, the helping/hindering distinction lights up. When that goal is ambiguous, results are noisier. So the underlying capacity may be real but more sensitive to design choices than the early studies suggested. The work has not been disproven; it has been complicated, which is what usually happens to interesting findings as they age.

Candid documentary snapshot of two Black toddlers, around 14 months old, both with short curly hair and wearing pastel cotton rompers, sitting opposite each other on a sunlit playmat. One is offering a wooden block to the other, who is reaching out with an open hand. Soft natural light, slight motion blur on the hands

What does this mean for parents in real life?

Take the headline at face value, and you might come away thinking your baby is silently rating you and your friends. The reality is gentler. The puppet-show experiments do not show that babies remember specific people as good or bad after a single bad day. They show that the basic machinery for “is this person helping or hurting another person” is online from the middle of the first year.

What that suggests, practically, is twofold. First, your baby is watching the way you treat other people, not just the way you treat them. Sibling interactions, kindness toward a partner, patience with a stranger on the phone: these are data points the baby’s social-evaluation system is taking in, even when they are too young to repeat what they saw. Second, you do not need to perform. There is no curriculum here. Ordinary, consistent, helpful behavior in a baby’s environment seems to be exactly the thing the baby’s brain is designed to notice.

It is also worth saying what the research does not say. It does not say babies have moral reasoning, in the philosophical sense. It does not say a 6-month-old is making judgments about character, fairness, or virtue. It says they have a preference, an early sorting mechanism, and that mechanism happens to point in the same direction adult morality eventually points. The grown-up moral life is built on top of that.^3,5

Overhead candid snapshot of a South Asian father in his mid-thirties with short black hair and a trimmed beard, sitting on a hardwood floor and reading a board book with bright illustrations to a 10-month-old baby with dark hair and golden-brown skin. The baby is pointing at a character on the page. Warm lamp light, cozy living room blur in the background

Common questions about babies and moral judgment

At what age do babies start preferring helpful people?

Robust effects have been reported as early as 6 months, with weaker but measurable preferences in 3-month-olds when researchers track looking time rather than reaching.^1,3

Are these preferences strong enough to predict behavior in real life?

Not yet. The puppet-show paradigm captures a brief, controlled judgment in a lab. There is no evidence it predicts how a specific baby will treat playmates years later. It is a window into a capacity, not a personality test.

Do babies understand intent or just outcome?

By around 8 months, infants weigh intent: a character who tried to help but failed is still preferred over one who tried to harm and missed. Younger infants rely more on outcome.⁵

Is this finding considered settled science?

The basic phenomenon has been replicated by multiple labs, but specific design features matter. The current consensus is that something real is going on, and that it is more sensitive to experimental setup than first reported.

Does culture change the result?

Cross-cultural data are still thin. Most published work on the Helper/Hinderer paradigm comes from Western lab samples. Whether the same preference shows up at the same ages worldwide is an open question that newer research is starting to address.

The takeaway, with the rough edges left on

The picture that comes out of two decades of work is not that your baby is a tiny judge with a gavel. It is quieter and, in a way, more interesting. From around the middle of the first year, a baby’s brain seems to come equipped with a small but real instrument for sorting social behavior into helpful and harmful, and that instrument starts running long before the baby can name what it is doing.

That instrument is rough. It can be fooled by experimental tweaks. It does not yet handle context the way an adult’s does. But it is there, and it points the same direction the rest of moral development eventually goes. Your little one is paying attention, and the people around them are part of the curriculum, whether anyone meant to enroll or not.

Sources

Hamlin JK, Wynn K, Bloom P. Social evaluation by preverbal infants. Nature. 2007. PubMed: 18033298
Hamlin JK, Wynn K, Bloom P. Three-month-olds show a negativity bias in their social evaluations. Developmental Science. 2010. PubMed: 20977563
Hamlin JK, Wynn K. Young infants prefer prosocial to antisocial others. Cognitive Development. 2011. PubMed: 21499550
Hamlin JK, Wynn K, Bloom P, Mahajan N. How infants and toddlers react to antisocial others. Proceedings of the National Academy of Sciences. 2011. PubMed: 22123953
Hamlin JK. Failed attempts to help and harm: intention versus outcome in preverbal infants’ social evaluations. Cognition. 2013. PubMed: 23811094