Learning

• 1. Introduction

  ›

  Learning begins at birth, and the evidence suggests it may start even earlier. Researchers have observed habituation in a fetus as early as 32 weeks into gestation. That finding indicates the central nervous system is already primed for memory before a child has taken a first breath. From there, learning continues until death, driven by ongoing interactions between people and their environment. It is the process of acquiring new understanding, knowledge, behavior, skills, values, attitudes, and preferences. Some of it arrives in an instant, like being burned by a hot stove. Most of it accumulates from repeated experiences over a lifetime. But here is a stranger truth. This capacity does not belong to humans alone. There is evidence for some kind of learning in other animals, in certain machines, and even in certain plants. So what counts as learning when a song bird, a protozoan, and a garden pea all qualify? Why would a behavior as costly as play ever evolve? And how can a single dog teach us the difference between a reflex and a lesson? Those are the questions the rest of this story will answer.

• 2. Learning Without Association

  ›

  A stuffed owl dropped into a cage of small song birds will, at first, terrify them. They react as though a real predator has arrived. Soon they react less, and then barely at all. This is habituation, a form of non-associative learning in which a response to a repeated stimulus diminishes. Move that same owl and reintroduce it, and the panic returns. The birds had learned to ignore one particular unmoving owl in one particular place, nothing more. Habituation is precise, and it is also widespread. It has been shown in essentially every species of animal, as well as the sensitive plant Mimosa pudica and the large protozoan Stentor coeruleus. The speed of the process depends on the stimulus. It happens faster for stimuli that occur at a high rate than for those that occur rarely. Crucially, habituation must be distinguished from extinction, which is an associative process where a response declines because a reward has stopped. Habituation has a mirror image called sensitization, in which a response grows stronger rather than weaker. Rub your arm continuously and a warm sensation builds, then eventually turns to pain. That pain is a progressively amplified synaptic response of the peripheral nerves, a warning that the stimulation is harmful. Sensitization is thought to underlie both adaptive and maladaptive learning, the same mechanism serving very different ends.

• 3. Pavlov's Bell And Watson's Albert

  ›

  Ivan Pavlov fed his dogs meat powder, and the dogs salivated. That salivation was a reflex, an unconditioned response to an unconditioned stimulus. Then Pavlov rang a bell before each meal. The first time, the bell meant nothing and the dogs stayed dry. After numerous pairings of bell and food, the dogs learned that the bell signaled food, and they salivated at the sound alone. The bell had become a conditioned stimulus, the salivation a conditioned response. This is classical conditioning, and it reaches far beyond dogs. It has been demonstrated in honeybees through the proboscis extension reflex, and recently even in garden pea plants. John B. Watson pushed this thinking into a new philosophy of science. He argued the introspective method was too subjective, and that psychology should limit itself to directly observable behaviors. In 1913 he published the article "Psychology as the Behaviorist Views", insisting laboratory studies served the field best. His most famous and most controversial experiment was "Little Albert", which showed how the learning of emotion could be accounted for through classical conditioning. Watson's behaviorism paved the way for B.F. Skinner's radical behaviorism, a direct contrast to Freud and other accounts built on introspection.

• 4. Reward, Punishment, And Imitation

  ›

  Operant conditioning rests on a simple premise. Living things seek pleasure and avoid pain, and behavior can be shaped by reward or punishment delivered at the right moment. That moment matters. The reward must follow immediately, in the small window called trace conditioning. Unlike classical conditioning, this approach shapes wanted behavior that requires conscious thought rather than relying on natural reflexes. The mechanics split into careful categories. Positive punishment adds something aversive, like a spanking. Negative punishment removes something desirable, like a child put in time out who loses the freedom to be with friends. Reinforcement works the other way, increasing a behavior. Negative reinforcement removes an undesirable thing, as when a dog learns to sit while a trainer scratches away an itch. Positive reinforcement adds a desirable thing, like a treat. Notably, punishment is held to be a poor tool for increasing wanted behavior, because it teaches avoidance of the punishment rather than of the unwanted act itself. Not all learning requires this scaffolding of consequences. Observational learning occurs simply through watching others, relying on a social model such as a parent, sibling, friend, or teacher. A related phenomenon was uncovered in 1935, when the Austrian zoologist Konrad Lorenz found that certain birds will follow and bond with an object if it makes sounds. That rapid, life-stage-bound attachment is called imprinting.

• 5. Why Play Pays

  ›

  Orcas have been seen playing with seals they have caught, and young cats bat at a ball of string. Both look like idle fun, yet both carry real danger. Play increases vulnerability to predators, risks injury and possibly infection, and burns energy. For a behavior that costly to survive evolution, the benefits must be significant. The clue lies in timing. Play appears mostly in younger animals, suggesting a link with learning, and the cat batting string is rehearsing the catch of prey. It shows up across a wide variety of vertebrates, though it is mostly limited to mammals and birds. In children, play becomes central to development. Through it they learn social skills like sharing and collaboration, emotional skills like managing anger, and the foundations of thinking and language. There are five recognized types. Sensorimotor play repeats an activity, roleplay begins around the age of three, rule-based play centers on prescribed codes of conduct, construction play involves building, and movement play is physical. These often intersect, and all of them generate thinking and problem-solving. Lev Vygotsky saw play as pivotal, the stage where a child first grasps rules and symbols. For Vygotsky it was the first form of learning language and communication itself.

• 6. The Many Schools Of Learning

  ›

  A child bitten by a dog who then fears all dogs has undergone episodic learning, a change in behavior caused by a single event. Such events are stored in episodic memory, which recalls experience embedded in time. Remembering the Grand Canyon from a recent visit is episodic. Knowing where the Grand Canyon is draws instead on semantic memory, which strips facts from their experiential context. Beyond these run a long roster of named approaches. Rote learning memorizes information for exact recall through repetition, used everywhere from mathematics to music to religion. Meaningful learning is its opposite, where a fact is understood through its relationship to other knowledge. Enculturation teaches the values of one's own culture, distinct from acculturation, which adopts the rules of a different one. Among the Mazahua people, collaborative practice rooted enculturation in nonverbal social experience, and Chillihuani girls in Peru described themselves as weaving constantly by following the adults. Formal learning carries goals, organization, and often a diploma. Informal learning is the happenstance of day-to-day life, like learning to look ahead while walking, requiring no enrollment and leading to no accreditation. Tangential learning shows yet another route. People self-educate when a topic appears inside something they already enjoy. The game designer James Portnow was the first to suggest games as a venue for it, pointing to the built-in encyclopedias of the Civilization series as a place gamers dig deeper into history.

• 7. What Helps And What Hinders

  ›

  Benjamin Bloom mapped learning into three domains, and a single game of chess touches all of them. Learning the rules is cognitive. Setting up and moving the pieces is psychomotor. Coming to love the game and appreciate its history is affective. The domains are not mutually exclusive. Beyond the domains, retention has its own levers. The spacing effect shows that lessons spread over time beat cramming, because cramming fights the forgetting curve. Paraphrasing material to oneself outperforms passive reading, and low-stakes quizzing produces the testing effect. Motivation sits underneath all of it. Intrinsic motivation, a learner's own curiosity, sustains learning more effectively than extrinsic rewards like grades. Rote drill has been criticized as "drill and kill" precisely because it can kill that curiosity. Conditions matter too. Malnutrition, fatigue, and poor health slow learning, as do bad ventilation, poor lighting, and crowded rooms that raise stress and break sightlines to the instructor. Deeper still, the molecular basis of learning appears to be dynamic changes in gene expression in brain neurons, introduced by epigenetic mechanisms like the methylation of neuronal DNA. Age plays its part as well. Children's brains contain more silent synapses, inactive until recruited for flexible learning, and neuroplasticity is heightened during critical periods of development. Yet some researchers, after putting late middle aged participants through university courses, suggest the apparent gap may owe more to time, support, and attitude than to inherent ability.

• 8. The Pea That Learned

  ›

  Monica Gagliano, an Australian professor of evolutionary ecology, set out to test whether a plant could be trained. Her subject was the garden pea, Pisum sativum, placed inside Y-shaped tubes. In training sessions, seedlings met light coming down one arm and a gently blowing fan down the same or opposite arm. The light was the unconditioned stimulus the plant naturally bends toward. The wind was the conditioned stimulus. In the testing phase, the seedlings faced the fan alone, with no light at all. The majority grew toward the arm where light had been predicted by the fan's position the day before, while a control group showed no preference. The mechanism remains unknown, though one hypothesis ties photoreception to mechano-perception through stretch-gated ion channels that let calcium flood the cell. The claim has drawn controversy, much of it over language rather than data. The author and journalist Michael Pollan, in his piece The Intelligent Plant, noted that critics do not doubt Gagliano's results so much as her use of the words "learning" and "cognition". The psychologist Charles Abrahmson points to a deeper problem, that researchers do not even share a consistent definition of those terms. That same ambiguity travels into a very different domain. Machine learning, a branch of artificial intelligence, builds systems that learn from data, such as one trained on email to separate spam from the rest. The same question that hovers over a pea in a Y-tube hovers there too, and a proposed direction for future research is simply surveying how scientists define cognition at all.