Information processing (psychology)

• 1. Introduction

  ›

  Information processing in psychology asks a deceptively simple question: what does the mind actually do when it thinks? The approach that emerged in the 1940s and 1950s, in the years after World War II, proposed a startling answer. Cognition, it suggested, is essentially computational in nature. The mind is the software; the brain is the hardware.

  That framing borrowed its language from the machines that were reshaping the postwar world. It brought with it a set of tools for studying thinking that had never existed before: measurable stages, testable models, and the idea that mental operations unfold in an orderly sequence that science can observe. From memory scanning experiments run in the 1960s to brain-imaging research in the twenty-first century, the approach has generated an expanding family of models and theories. Each one tries to answer a version of the same question: how does information get in, how is it held, and how is it used?

  At the heart of the framework lies a deceptively humble organ. The central executive of working memory has been compared to a secretary of the brain. It decides what needs attention and how to respond. The chapters ahead follow that secretary through the full sweep of what information processing has revealed about human cognition.

• 2. The Two Directions of Processing

  ›

  Information processing can run in two directions. Vertical processing moves up or down a hierarchy; horizontal processing spreads across a network. Either kind can be centralized in one place or distributed across many.

  In the mid-1980s, the distributed horizontal approach gained wide attention under the name connectionism. A connectionist network is made up of different nodes, and knowledge is not stored in any single node. Instead, meaning emerges from a combination of differently activated nodes working in concert. The mechanism that links them is called a priming effect: a prime node activates a connected node, spreading activation across the network. This is different from earlier semantic network models, where a single node carried a specific meaning. In connectionism, the knowledge lives in the pattern of activation, not in any one location.

  The distinction between these two styles of processing matters because it shapes predictions about how the mind can fail. A centralized system is fragile at its center; a distributed one can lose nodes without losing the whole pattern. The emergence of connectionism in the 1980s opened new lines of inquiry into memory, language, and perception that the earlier, more hierarchical models had not anticipated.

• 3. Sternberg's Components of Thinking

  ›

  Psychologist Saul Sternberg introduced the concept of high-speed memory scanning in the 1960s, and it was one of the first demonstrations that mental operations occur in orderly, measurable stages. In a series of experiments published in 1966 and 1969, Sternberg asked participants to memorize small sets of numbers and then judge whether a test number had been part of the set. Reaction time increased in a straight line as the number of items grew. That result pointed to a process of serial exhaustive search: people appear to check each item in memory one by one, running through the entire set even after finding a match. Sternberg continued to defend this model as recently as 2016, arguing for its continued importance in understanding how the mind processes and retrieves information.

  Robert Sternberg's later triarchic theory of intelligence extended this computational view into a broader framework. He proposed three components: creative, analytical, and practical abilities. Creative ability generates new ideas; analytical ability evaluates whether an idea is sound; and practical ability implements ideas and persuades others of their value. Within this model, information processing sits at the center of cognition and is itself divided into three levels. Meta components handle planning and evaluation, operating as higher-order executive functions. Performance components carry out the instructions of the meta components. Knowledge-acquisition components learn how to approach problems in the first place.

  A working art project illustrates how these levels interact. A decision about what to draw, a plan, and a rough sketch all belong to meta component processing. The act of drawing itself is performance. Learning to draw better over time is knowledge acquisition. The three levels nest inside each other rather than running in a simple sequence.

• 4. Working Memory's Architecture

  ›

  The Atkinson-Shiffrin memory model, also called the multi-store model, describes three stages that information must pass through to be firmly planted in memory: sensory memory, short-term memory, and long-term memory. Working memory sits at the center of this architecture and has its own internal structure.

  At the top is the central executive, which allocates attention and directs responses. Below it run three distinct subsystems. The phonological loop handles language: it takes in sounds, converts them to a form the brain can rehearse subvocally, and holds them long enough to be stored. The visuospatial sketchpad stores visual images; its capacity is brief but allows the mind to manipulate what it sees. The episodic buffer does something neither of the others can do alone: it integrates material from the phonological loop and the visuospatial sketchpad with information already in long-term memory, producing what researchers describe as a unitary episodic representation.

  All of this begins at the sensory register, which takes in information through all five senses simultaneously. The capacity of the sensory register is large, but its duration is very short, lasting roughly 1-3 seconds. Linden's research showed that short-term memory's capacity was initially estimated at "seven plus or minus two" items, a number consistent with the average digit span of healthy adults. Those items can be held for about 5-20 seconds; without rehearsal, they decay. Crucially, retention improves when items are grouped into chunks using perceptual or conceptual associations. Long-term memory, by contrast, has a potentially unlimited capacity and a duration that is effectively indefinite, though retrieval can sometimes feel like a fact sitting on the tip of the tongue.

• 5. Visual Memory and Personal Bias

  ›

  DeStefano, Vul, and Brady published research in 2025 that refined the picture of how visual information is held in working memory. Their experiments asked participants to remember and then reproduce colors from memory. What emerged was a pattern of consistent personal biases: each individual showed a pull toward certain hues, which the researchers called attractor biases.

  These attractors are not random noise. They reflect individual perception and the weight of past experience. The finding shifted the information processing framework in a meaningful direction. Earlier models focused largely on universal mechanisms: stages that apply to everyone in the same way. This work showed that stable internal cognitive tendencies also shape how the brain handles incoming information. Two people shown the same color may encode and retrieve slightly different versions of it, each nudged by their own attractor.

  The implication is that information processing is neither purely stimulus-driven nor purely idiosyncratic. It involves a negotiation between what arrives from the outside world and what the individual brain brings to the encounter. This personal dimension had been harder to study in earlier decades, before methods existed to detect such fine-grained individual differences in memory reproduction.

• 6. Piaget and Developmental Stages

  ›

  Jean Piaget approached information processing from a developmental angle. He proposed four distinct stages, each tied to an age range and characterized by a specific type of thinking. From birth to age 2, the sensorimotor stage sees infants relying on their senses and responding with reflexes. Between ages 2 and 6, the preoperational stage brings language and imitation, but children at this stage cannot yet take another person's point of view. From ages 6 to 11, the concrete operational stage introduces logic and the ability to consider multiple factors when solving a problem. From age 11 onward, the formal operational stage enables abstract thinking, argumentation, and the generation of counter-arguments.

  Adolescence is where some of the most significant biological changes in cognitive processing occur. Two brain regions are central to this period. The prefrontal cortex, which is active during planning, goal generation, strategic thinking, and metacognition (thinking about thinking), does not reach full development until early adulthood. The limbic system, which regulates emotion and modulates reward sensitivity through neurotransmitters including dopamine, also undergoes important changes during adolescence.

  The interaction between these two regions helps explain why adolescent decision-making looks the way it does. The limbic system's reward sensitivity can run ahead of the prefrontal cortex's capacity for risk evaluation, a gap that narrows as the cortex completes its development. Adults, Piaget's framework suggests, are better equipped than children or adolescents to plan, comprehend abstract concepts, and evaluate risks and benefits.

• 7. Beyond Psychology

  ›

  The phrase "information processing" extends well beyond psychology. In computing, it refers broadly to the use of algorithms to transform data, and in that sense it describes the defining activity of computers themselves. The International Federation for Information Processing, known as IFIP, is one broad professional organization that takes this wider definition as its scope.

  In computing contexts, information processing is effectively synonymous with data processing or computation, though the term carries a more general connotation. The parallel to the psychological version is not incidental. The postwar computational metaphor that gave rise to the psychological approach also shaped how engineers and mathematicians talked about machines. Both fields describe systems that take in signals, operate on them through structured stages, and produce outputs. Whether the substrate is silicon or neurons, the vocabulary converged.

  The philosophical counterpart to the psychological approach is the computational theory of mind, which holds that mental states are themselves a kind of computation. Cognitivism in psychology and functionalism in philosophy occupy adjacent territory. All three positions share the core idea that the details of the physical substrate matter less than the pattern of operations being performed, a claim that the connectionist models of the 1980s would test and partly complicate.