Abstract
Cognitive psychology is the branch of psychology, and a core part of cognitive science, that studies attention, perception, memory, language and thought as the processing of information. Its first comprehensive textbook appeared in 1967, after a mid-century turn away from behaviorism that one participant's personal account dates to the 1950s and a historical review describes as often slow and piecemeal. Classic work showed that attention can shift without eye movements, that immediate memory holds only a few chunks, and that judgment leans on heuristics whose costs and benefits remain contested. Grounded cognition, predictive processing and the critique of WEIRD samples challenge parts of the framework, and a 100-study replication project found replication effects about half the size of the originals. Current work turns the field's tasks on large language models.
Keywords: cognitive psychology, information processing, cognitive revolution, working memory, attention
Cognitive psychology investigates how people attend, perceive, remember, use language, judge and think, and it treats each of these as a form of information processing: input from the senses is taken up, recoded, held and put to use. The field took its first comprehensive textbook from Ulric Neisser's Cognitive Psychology, published in 1967 (Neisser, 2014).
This page sets out what the field studies, how it emerged from and against behaviorism, the information-processing framework and its classic findings on attention, memory and judgment, the main computational models, the serious challenges the framework now faces, the replication debate, and the lines of work that are currently most active. Each topic is treated briefly, with a link to the site's fuller page where one exists.
Key Takeaways
- Cognitive psychology treats mental life as the transformation of stimulus information, a framing set out in the field's first comprehensive textbook in 1967.
- One personal account dates the shift away from behaviorism to the 1950s, while a historical review describes it as often slow and piecemeal and largely a return to older attitudes.
- A theoretical review of earlier studies put the central capacity of short-term memory at about four chunks rather than seven, a limit visible only under constrained conditions.
- Heuristics were described in a 1974 paper as sources of systematic, predictable error, yet the same authors called them usually effective, and a later review argued that ignoring information can raise accuracy, for instance when predictability is low or samples are small.
- In a project that repeated 100 published studies, 36% of replications reached significance against 97% of the originals, and no single standard for replication success was agreed.
What Cognitive Psychology Is
Following the plan of Neisser's textbook, cognitive psychology studies what happens to stimulus information once it reaches an observer: how it is transformed and reconstructed, and how it figures in thought and memory (Neisser, 2014). The subject matter is therefore the set of mental processes that lie between what is sensed and what is done, from perception and attention to memory, problem solving and decision making.
Within the classification used by the National Library of Medicine, cognitive psychology sits under two broader headings: psychology, of which it is a branch, and cognitive science. The second placement reflects history. George Miller, writing from personal experience rather than as a systematic historian, dated the birth of cognitive science to the 1950s and described it as the product of several disciplines working at once (Miller, 2003). Cognitive psychology is the experimental, behavioural part of that enterprise, and it now shares much of its ground with cognitive neuroscience.
A useful statement of what counts as a cognitive explanation comes from David Marr's posthumously published book on vision, which separated three levels of analysis: the computational level (what problem a system solves), the algorithmic level (the representations and procedures it uses) and the hardware implementation level (how those procedures are physically realised); Marr applied the scheme to vision as a series of stages leading from the image to a three-dimensional description of objects, and the 2010 reissue notes that the main problems that occupied him remain open problems in the study of perception (Marr, 2010).
The Cognitive Revolution
For roughly three decades, from 1920 to 1950, behaviorism dominated American psychology, until it was replaced because it could not address central issues in human psychology (Mandler, 2002). The usual story of how that dominance ended has two versions.
The first version stresses rupture. Noam Chomsky's 1959 review of B. F. Skinner's Verbal Behavior argued that behaviorist and empiricist accounts could not explain verbal behavior or higher mental processes. Read with its retrospective preface, which the author added later, the review called Skinner's study careful and thorough, and it treated that very care as what made the work a reductio ad absurdum of behaviorist assumptions: in Chomsky's view the details could be improved, but the framework could not (Chomsky, 1959). Miller's own recollection places the turn in the 1950s and casts it as a convergence of several fields on the study of mind, though it is explicitly a personal account rather than a systematic history (Miller, 2003).
The second version stresses continuity. George Mandler's historical review described the change as often slow and piecemeal, and argued that much of it was a return to attitudes toward mind that had been common before behaviorism displaced them; his title marks the point by writing the event as a cognitive (r)evolution (Mandler, 2002). The two readings differ less on what happened than on how abrupt it was, and the cited material does not settle which is the better description.
The Information-Processing Approach
The framework that emerged treats the mind as a system that takes in stimulus information and transforms, reconstructs, stores and uses it, and Neisser's 1967 textbook gave it its first comprehensive statement (Neisser, 2014). Much early work asked where the bottlenecks in such a system lie. An early book-length treatment of such questions is Broadbent's Perception and Communication (Broadbent, 1958). The question of whether selection happens early or late was later recast by Nilli Lavie and colleagues' load theory of attention, which proposes two mechanisms of selective attention: in their series of experiments, distractors interfered less when perceptual load was high, but more when working memory or the coordination of two tasks was heavily loaded (Lavie et al., 2004).
George Sperling's tachistoscopic experiments, a condensed doctoral thesis, examined how much of a brief visual presentation an observer could report, and whether the limit lay in reporting or in memory (Sperling, 1960); that line of work is the background to the site's page on sensory memory. Richard Atkinson and Richard Shiffrin then proposed a system of human memory together with the control processes that govern it (Atkinson & Shiffrin, 1968), the account usually called the multi-store model. Its standing today is revised: later theory moved from a single short-term store toward working memory, the subject of Baddeley and Hitch's 1974 chapter (Baddeley & Hitch, 1974), and current models of short-term and working memory are measured against an agreed set of benchmark findings (Oberauer et al., 2018).
Fergus Craik and Robert Lockhart offered levels of processing as an alternative framework for memory research (Craik & Lockhart, 1972). Thirty years on, Craik returned to the framework in a retrospective review that defended depth of processing as an account of memory (Craik, 2002).
Thinking itself was studied in the same terms. Newell and Simon's 1972 book Human Problem Solving took human problem solving as its subject, and the site's page on problem solving covers that line of work (Newell & Simon, 1972).
The information-processing approach as a whole is best described as revised rather than overturned. It remains the working vocabulary of the field, but it has been challenged by accounts that ground cognition in the body (Barsalou, 2008; Wilson, 2002), by the view of the brain as a prediction machine (Clark, 2013), and by the argument that brains, like large neural networks, may fit the world directly rather than through compact theories (Hasson et al., 2020). The section on challenges below takes these up.
Figure 1
The Information-Processing Framework as a Flow of Stages
Attention and Perception
Attention can move without the eyes moving. Michael Posner reviewed a series of experiments built around a visual-space attention-cuing task, run with alert monkeys, brain-injured patients and normal human observers, and concluded that a covert attentional mechanism shifts across the visual field while the eyes stay at fixation, a system whose relationship to saccadic eye movements the comparison examined, with gains in detection efficiency that can be traced to where attention has been directed (Posner, 1980). A parallel form of orienting in semantic memory was only explored in that work, not established.
Anne Treisman and Garry Gelade proposed a feature-integration theory of attention (Treisman & Gelade, 1980); the site's page on feature integration theory covers it in detail. A later account of visual search, Wolfe's Guided Search 2.0, was worked out as a computer simulation in which features registered in parallel at a preattentive stage steer where a limited-capacity stage is deployed, and its author acknowledged shortcomings and issues still unresolved (Wolfe, 1994).
A sharper dispute concerns whether beliefs, desires and expectations change what is seen, a question known as cognitive penetrability. In a critical review of the literature on top-down effects, Chaz Firestone and Brian Scholl found no compelling evidence that cognition alters perception itself. They set themselves against hundreds of published studies, argued that a handful of recurring pitfalls accounted for the apparent effects, and offered a checklist for future studies, while granting that substantive top-down effects might yet be demonstrated (Firestone & Scholl, 2016). The opposing case is put by Lupyan, whose paper on cognitive penetrability in an age of prediction argues that predictive systems are penetrable systems (Lupyan, 2015). The question bears directly on visual perception and on selective attention, and it remains open.
Move Attention Without Moving the Eyes
Spatial Cueing: Benefits and Costs
Choose a trial type. The cue lights a location; the target (X) then appears at the cued location (valid), at the other location (invalid), or after a cue that points nowhere (neutral).
Memory and Its Limits
Immediate memory is small. In his essay on the magical number seven, Miller surveyed the ceilings people meet in absolute judgment and in immediate memory span, and proposed that recoding material into chunks lets them stretch those ceilings (Miller, 1956). Miller was cautious about the number itself: he allowed that the recurring seven might be a coincidence and withheld judgment on what it meant (Miller, 1956). Chunking has remained central to the field since.
Nelson Cowan's theoretical review of prior studies reconsidered the figure and put the central capacity of short-term memory at about four chunks rather than 7 ± 2 (Cowan, 2001). The estimate comes with conditions. The pure capacity limit can be observed only under constrained conditions, chunking is often not measured directly, compound estimates run above the pure limit, and individuals vary more widely than the average suggests (Cowan, 2001). A rival account doubts that capacity is a fixed count of slots, whether seven or four: a review by Ma and colleagues set that view against one in which working memory is a limited but flexible resource, spread across items and bearing on the quality of each representation, although the neural evidence for the resource view was only beginning to emerge (Ma et al., 2014).
Over the same period, theorists moved from a single short-term store to working memory, the topic of Baddeley and Hitch's 1974 chapter (Baddeley & Hitch, 1974). The working memory model is treated here as supported: Baddeley's 2012 review presented the overall multicomponent framework as relatively stable, as distinct from the attempts to build more specific models within it, and closed with speculations on further developments (Baddeley, 2012). A rival account treats working memory as simply the activated part of long-term memory. Looking back over 50 years of the modal and multicomponent models, Baddeley, Hitch and Allen argued on theoretical grounds for keeping a separate short-term store and doubted that activated long-term memory could take its place (Baddeley et al., 2019). A 2018 paper by 16 authors, informed by an expert survey, assembled benchmark findings that any theory of short-term and working memory should explain; the selection rested on the authors' consensus, aimed at a theory-neutral description, and covered only a subset of the findings in the field (Oberauer et al., 2018).
Recode the List
Chunking: Twelve Letters, Fewer Units
Move the slider to group the same twelve letters into larger chunks. The memory load is the number of chunks, not the number of letters.
Judgment and Decision Making
The standing of heuristics-and-biases research is contested, and the dispute is about what heuristics are for. Amos Tversky and Daniel Kahneman's theory paper identified three heuristics used in judgment under uncertainty, namely representativeness, availability, and adjustment from an anchor, and showed that they produce systematic, predictable errors (Tversky & Kahneman, 1974). In the same theory paper, the authors also judged these heuristics to be economical and, most of the time, effective (Tversky & Kahneman, 1974).
Gerd Gigerenzer and Wolfgang Gaissmaier's review approached heuristics as adaptive tools and surveyed formal models of heuristic inference. It reported less-is-more effects, in which ignoring part of the available information improves accuracy, and tied those gains to environments of low predictability and to small samples (Gigerenzer & Gaissmaier, 2011). The authors acknowledged that a systematic theory of heuristics does not yet exist (Gigerenzer & Gaissmaier, 2011). Each programme therefore concedes part of the other's point, and the balance between error and adaptation remains unsettled; the work connects to decision making and to bounded rationality.
Models of the Mind
Computational models turned verbal theories into programs whose behaviour could be checked against data. James McClelland and David Rumelhart's interactive activation model, a computer simulation with detectors for features, letters and words and with feedback from the word level to the letter level, accounted for the advantage of letters seen in words and the facilitation produced by pseudowords (McClelland & Rumelhart, 1981). The idea is supported in revised form. Revisiting it in 2014, McClelland and colleagues reviewed the broader interactive activation hypothesis, in which activation spreads in both directions until the levels of processing settle into mutual constraint satisfaction, and drew support for it from computational, behavioural and neuroscience work; they also granted that the early models departed from optimal Bayesian inference and fitted some human data poorly, and they took up the continuing controversy over interactive versus Bayesian accounts (McClelland et al., 2014).
ACT-R, set out by John Anderson and colleagues as an integrated theory of the mind, combines perceptual-motor modules, a goal module and a declarative memory module, which communicate through buffers holding chunks and are coordinated by production rules and subsymbolic processes (Anderson et al., 2004). The architecture remains active: a 2019 review by Frank Ritter and colleagues described ACT-R as a hybrid cognitive architecture used to model and predict human behaviour, and named emotions, physiology and usability among the areas still to be developed (Ritter et al., 2019).
Building and fitting such models takes discipline. For readers who want to build such models, a tutorial by Robert Wilson and Anne Collins sets out ten simple rules for fitting computational models to behavioral data, taking the reader from simulating a model through estimating its parameters to comparing it with rivals (Wilson & Collins, 2019). Their worked examples were limited to reinforcement learning and to the simplest techniques, although they expected most rules to carry over to advanced methods, and they stressed that modeling cannot replace good experiments (Wilson & Collins, 2019).
Table 1
Major Theories and Models in Cognitive Psychology
| Theory | Core claim | Status | Key source |
|---|---|---|---|
| Information-processing approach | Cognition transforms and reconstructs stimulus information in perception, thought and memory | Challenged (Hasson et al., 2020) | Neisser, 2014 |
| Multi-store model | A proposed system of human memory and its control processes | Revised (Baddeley et al., 2019) | Atkinson & Shiffrin, 1968 |
| Working memory model | Short-term retention understood as working memory | Supported (Baddeley, 2012) | Baddeley & Hitch, 1974 |
| Levels of processing | A framework for memory research | Contested (Craik, 2002) | Craik & Lockhart, 1972 |
| Feature integration theory | A feature-integration theory of attention | Revised in later models (Wolfe, 1994) | Treisman & Gelade, 1980 |
| Interactive activation model | Interaction between letter and word levels explains the word advantage in letter perception | Revised (McClelland et al., 2014) | McClelland & Rumelhart, 1981 |
| ACT-R | Specialised modules integrated through buffers and coordinated by production rules | Supported (Ritter et al., 2019) | Anderson et al., 2004 |
| Heuristics and biases | Three judgment heuristics, usually effective, produce systematic, predictable errors | Contested (Gigerenzer & Gaissmaier, 2011) | Tversky & Kahneman, 1974 |
| Grounded cognition | Cognition rests on modal simulations, bodily states and situated action rather than amodal symbols | Contested (Wilson, 2002; Mahon & Caramazza, 2008) | Barsalou, 2008 |
| Predictive processing | The brain is a hierarchical prediction machine that minimises prediction error | Contested (Bowers & Davis, 2012) | Clark, 2013 |
Note. The Status column gives each theory's present standing in the field, and the citation beside it names the work on which that standing is based.
Challenges to the Framework
The most direct challenge comes from grounded cognition, which is contested. Lawrence Barsalou's review argued that cognition rests on modal simulations, bodily states and situated action rather than on amodal symbols, while acknowledging open theoretical, empirical and methodological issues and noting common misperceptions of the view (Barsalou, 2008). Margaret Wilson pulled apart six distinct claims grouped under embodied cognition and judged their support uneven: treating the environment as part of the cognitive system was problematic, four of the claims held in part, and the idea that off-line cognition is body based, though it had received the least study, might be the strongest of them (Wilson, 2002). A sharper critique came from Mahon and Caramazza, who concluded that the evidence does not support the embodiment conclusion: strong embodiment runs into neuropsychological data, and the sensory-motor activation findings offered in its favour can be explained just as well by disembodied accounts (Mahon & Caramazza, 2008).
Predictive processing, also contested, recasts the brain as a hierarchical prediction machine. In a target article, Andy Clark examined hierarchical predictive coding, in which a generative model issues top-down predictions and the system works to reduce prediction error, weighted by precision. He framed the article as a critical examination rather than a proof and set out evidential, methodological and conceptual pitfalls (Clark, 2013). Bayesian accounts more broadly have drawn a related critique: Bowers and Davis judged theories that cast cognition as optimal Bayesian inference to be only weakly supported, found the neuroscience evidence for them weaker still, and argued that non-Bayesian alternatives deserve a hearing (Bowers & Davis, 2012). The site's page on predictive coding treats the mechanism further.
A third challenge questions the explanatory style of the field itself. Uri Hasson and colleagues proposed that the brain may work like over-parameterized artificial neural networks, which fit their data directly through an optimisation resembling evolution, and they presented this as a challenge to psychology's theoretical assumptions; the brain analogy was posed as a question rather than asserted (Hasson et al., 2020).
The last challenge concerns who is studied. After comparing data from many populations, Joseph Henrich, Steven Heine and Ara Norenzayan judged that the samples behavioural science leans on, drawn from Western, Educated, Industrialized, Rich and Democratic (WEIRD) societies, do not stand for humanity at large and often sit at the extremes of the human range (Henrich et al., 2010). They argued that there are no a priori grounds for assuming a finding is universal, though their evidence spanned selected domains only (Henrich et al., 2010).
Replication and Method
The field's results do not all replicate at their original strength. The Open Science Collaboration repeated 100 experimental and correlational studies from three journals. Of the original studies, 97% had reported significant results (P < .05), against 36% of the replications, and the mean effect size fell from Mr = 0.403 (SD = 0.188) in the originals to Mr = 0.197 (SD = 0.257) in the replications (Open Science Collaboration, 2015). The project used several indicators of success and held that no single standard defines one; its five indicators were not exhaustive, and its analyses of what predicted replication were correlational only (Open Science Collaboration, 2015).
Brian Nosek and colleagues' review treated the replication failures as a spur to self-correction across psychological science, separating replicability, robustness and reproducibility. It also noted that replication remains uncommon and often misunderstood, and that researchers face disincentives to conduct it (Nosek et al., 2022).
Compare Originals and Replications
How Often Did Published Findings Replicate?
Switch between two of the project's indicators.
Worked Example
The arithmetic below is illustrative and comes from no study. Suppose a list of twelve letters is held one letter at a time; it occupies 12 units. Grouped into threes, the same letters form 12 / 3 = 4 chunks, which sits inside the central estimate of about four chunks (Cowan, 2001). Grouped into pairs, they form 12 / 2 = 6 chunks, which is within Miller's seven but above four (Miller, 1956). The number of letters never changes; only the number of units to be held does, which is why chunking matters for any statement about capacity.
The replication figures can be read the same way. The two means are 0.403 for the original studies and 0.197 for the replications (Open Science Collaboration, 2015). Dividing one by the other gives 0.197 / 0.403 = 0.49, so the average replication effect was about half the size of the average original effect; the division is simple arithmetic, not a figure reported by the project.
The ChunkingCapacityDemo reproduces the chunk counts above, and the ReplicationDemo reproduces the two means and their ratio. The response times shown in the AttentionCueingDemo are invented for display and are not data from any study.
Discussion
Several things have held. Capacity limits are real, even though their size has been revised: Miller's limits on immediate memory and Cowan's estimate of about four chunks agree that very little can be held at once, yet whether capacity should be counted in items at all is itself disputed by the view of working memory as a flexible resource (Miller, 1956; Cowan, 2001; Ma et al., 2014). Computational modelling of behaviour has been described as having revolutionized psychology, and a tutorial sets out rules for using it with care, with the caveat that modelling cannot stand in for a good experiment (Wilson & Collins, 2019).
Other ideas were revised. The single short-term store gave way to working memory, now assessed against consensus benchmarks (Oberauer et al., 2018); the limit near seven was lowered; and the reading of heuristics as primarily sources of error has had to accommodate evidence that ignoring information can improve accuracy under some conditions (Gigerenzer & Gaissmaier, 2011).
Three questions remain contested. Grounded and embodied accounts are supported unevenly across their separate claims (Wilson, 2002); predictive processing is a programme with acknowledged pitfalls rather than a demonstrated account (Clark, 2013); and whether cognition penetrates perception turns on how well individual studies avoid known pitfalls (Firestone & Scholl, 2016). Settling these would take designs that pit the rival accounts against each other directly, preregistered and replicated; the cited sources do not show that such tests have been completed.
The replication findings bear on all of this. When replications reach significance far less often than originals and their mean effect is about half as large (Open Science Collaboration, 2015), the results that deserve most weight are those that have been repeated, and replication efforts have been credited with raising awareness that replication contributes to self-correction (Nosek et al., 2022).
Current Directions
One active line uses the field's own tasks to study artificial systems. Marcel Binz and Eric Schulz ran behavioral experiments on GPT-3 Davinci, a language model with 175 billion parameters, treating it as a simulated participant on tasks including the Linda problem, the Cognitive Reflection Test, the Wason card selection task, a multiarmed bandit horizon task, the two-step task and a Blicket experiment. The model did well on the bandit task but poorly on causal reasoning, failing the causal-chain condition; it showed no directed exploration, fell below humans on gamble decisions, and could be derailed by small changes to a prompt. The authors cautioned that many of the classic vignettes might have been part of its training set, which makes those results difficult to interpret, and their study probed one particular large language model (Binz & Schulz, 2023).
A related argument holds that machines will learn and think like people only if they build causal models of the world, draw on intuitive physics and intuitive psychology, and exploit compositionality and learning-to-learn. Brenden Lake and colleagues presented this as a position paper that builds on neural networks rather than dismissing them, and as an argument rather than an empirical test (Lake et al., 2017). It stands in tension with the direct-fit view, which questions whether compact models of this kind are what brains use at all (Hasson et al., 2020).
Within the laboratory, recent work includes contributions on method and on agreed targets: rules for fitting and comparing computational models (Wilson & Collins, 2019), the push for replication and self-correction (Nosek et al., 2022), benchmark findings for theories of working memory (Oberauer et al., 2018), and a checklist for any future claim that cognition shapes perception (Firestone & Scholl, 2016).
Common Misconceptions
- Short-term memory holds seven items.
- Miller's famous seven described limits on immediate memory that chunking can stretch, and Miller himself allowed that the number might be a coincidence (Miller, 1956). A later review put the central capacity at about four chunks, observable only under constrained conditions (Cowan, 2001).
- Heuristics are simply errors in thinking.
- The paper that made heuristics famous for their biases also called them economical and usually effective (Tversky & Kahneman, 1974). A later review argued that heuristics can be adaptive and that ignoring information can raise accuracy, for instance where predictability is low or samples are small (Gigerenzer & Gaissmaier, 2011).
- Findings from student samples describe all humans.
- Participants from Western, Educated, Industrialized, Rich and Democratic societies are often outliers rather than typical humans, and there are no a priori grounds for assuming that a finding from them is universal, although the comparative evidence covers selected domains only (Henrich et al., 2010).
Glossary
- ACT-R.
- An integrated cognitive architecture in which perceptual-motor, goal and declarative memory modules communicate through buffers and are coordinated by production rules.
- Chunk.
- A unit formed by recoding several items into one; Cowan expressed the capacity of short-term memory in chunks.
- Cognitive penetrability.
- The proposed capacity of beliefs, goals or expectations to change perception itself; a critical review found no compelling evidence for it while leaving future effects possible.
- Cognitive revolution.
- The mid-twentieth-century turn from behaviorism to the study of mind, dated to the 1950s in one personal account and described in a historical review as often slow and piecemeal and partly a return to earlier attitudes.
- Covert orienting.
- A shift of attention across the visual field made without a saccadic eye movement.
- Embodied cognition.
- A family of claims that cognition is situated, shaped by the body and by action, and partly off-loaded onto the environment; their support is uneven.
- Grounded cognition.
- The contested view that cognition rests on modal simulations, bodily states and situated action rather than on amodal symbols.
- Heuristic.
- A simple judgment strategy, usually effective, that can produce systematic errors and, under low predictability or small samples, can outperform strategies that use more information.
- Information processing.
- The framework that treats cognition as the transformation, storage and use of stimulus information.
- Interactive activation model.
- A simulation of letter perception in which feature, letter and word detectors interact, with feedback to the letter level, accounting for the word advantage.
- Levels of analysis.
- Marr's three ways of describing an information-processing system: computational, algorithmic and hardware implementation.
- Predictive processing.
- The contested view that the brain is a hierarchical prediction machine that uses a generative model to reduce prediction error.
- Replication.
- Repeating a study to test whether its result holds; in psychology it remains uncommon and faces disincentives.
- Short-term memory.
- Brief storage of a small amount of information, with a central capacity estimated at about four chunks under constrained conditions.
- WEIRD samples.
- Research participants from Western, Educated, Industrialized, Rich and Democratic societies, argued to be unrepresentative of humans and often outliers, on evidence from selected domains.
- Working memory.
- The system that took the place of a single short-term store in memory theory, now assessed against consensus benchmark findings.
Key Researchers
Alan Baddeley
(b. 1934). Emeritus Professor of Psychology at the University of York; with Graham Hitch he proposed the multicomponent model of working memory. ORCID - Wikidata - Wikipedia - Faculty Page
Nelson Cowan
Curators' Distinguished Professor Emeritus of Psychological Sciences at the University of Missouri; he argued that the central capacity of short-term memory is about four chunks. Wikidata - Wikipedia - Faculty Page
Daniel Kahneman
(1934–2024). Psychologist at Princeton University who, with Amos Tversky, described the heuristics people use in judgment under uncertainty and the biases they produce. Wikidata - Wikipedia
Brenden M. Lake
Associate Professor of Computer Science and Psychology at Princeton University; first author of an argument that machines that learn and think like people need causal models, intuitive theories and compositionality. Faculty Page
George A. Miller
(1920–2012). Psychologist at Princeton University, earlier at Harvard, whose paper on the magical number seven described limits on immediate memory and the role of chunking. Wikidata - Wikipedia
Ulric Neisser
(1928–2012). Psychologist at Cornell University whose 1967 book gave the field its first comprehensive textbook. Wikidata - Wikipedia
Michael I. Posner
(b. 1936). Professor Emeritus of Psychology at the University of Oregon; his cueing experiments studied the covert orienting of attention. Wikidata - Wikipedia
Anne Treisman
(1935–2018). Psychologist at Princeton University who, with Garry Gelade, proposed the feature-integration theory of attention. Wikidata - Wikipedia
Frequently Asked Questions
What is cognitive psychology?
Cognitive psychology is the branch of psychology that studies how stimulus information is transformed, reconstructed and used in perception, thought and memory, and it received its first comprehensive textbook in 1967 (Neisser, 2014).
What was the cognitive revolution?
It was the shift from behaviorism, which dominated American psychology from about 1920 to 1950, toward the study of mind; one historical review argues that the change was often slow and piecemeal and partly a return to attitudes that behaviorism had displaced (Mandler, 2002).
What is the information-processing approach?
It is the framework that treats the mind as a system that takes in stimulus information and transforms, stores and uses it, a view that was first set out comprehensively in a 1967 textbook (Neisser, 2014).
How much can short-term memory hold?
A theoretical review of earlier studies estimated its central capacity at about four chunks rather than seven, a limit that can be observed only under constrained conditions (Cowan, 2001).
What are heuristics?
Heuristics are simple strategies for judgment under uncertainty, such as representativeness, availability and adjustment from an anchor, which are usually effective but can lead to systematic, predictable errors (Tversky & Kahneman, 1974).
What does embodied cognition claim?
Embodied cognition is a set of separate claims about the role of the body, action and environment in thought, and a theoretical analysis of six such claims found their support to be uneven (Wilson, 2002).
Do cognitive psychology findings replicate?
Not always at full strength: when 100 published psychology studies were repeated, 36% of replications were significant against 97% of the originals, and the mean effect size roughly halved (Open Science Collaboration, 2015).
How does cognitive psychology relate to artificial intelligence?
Researchers now apply its classic tasks to language models; GPT-3 did well on a bandit task but poorly on causal reasoning, in procedurally generated tasks designed so that they could not have been in its training data (Binz & Schulz, 2023).
References
Anderson, J. R., Bothell, D., Byrne, M. D., Douglass, S., Lebiere, C., & Qin, Y. (2004). An integrated theory of the mind. Psychological Review, 111(4), 1036-1060. https://doi.org/10.1037/0033-295X.111.4.1036
Atkinson, R. C., & Shiffrin, R. M. (1968). Human memory: A proposed system and its control processes. In K. W. Spence & J. T. Spence (Eds.), Psychology of learning and motivation (Vol. 2, pp. 89-195). Academic Press. https://doi.org/10.1016/S0079-7421(08)60422-3
Baddeley, A. (2012). Working memory: Theories, models, and controversies. Annual Review of Psychology, 63, 1-29. https://doi.org/10.1146/annurev-psych-120710-100422
Baddeley, A. D., & Hitch, G. (1974). Working memory. Psychology of Learning and Motivation, 8, 47-89. https://doi.org/10.1016/S0079-7421(08)60452-1
Baddeley, A. D., Hitch, G. J., & Allen, R. J. (2019). From short-term store to multicomponent working memory: The role of the modal model. Memory & Cognition, 47(4), 575-588. https://doi.org/10.3758/s13421-018-0878-5
Barsalou, L. W. (2008). Grounded cognition. Annual Review of Psychology, 59, 617-645. https://doi.org/10.1146/annurev.psych.59.103006.093639
Binz, M., & Schulz, E. (2023). Using cognitive psychology to understand GPT-3. Proceedings of the National Academy of Sciences, 120(6), e2218523120. https://doi.org/10.1073/pnas.2218523120
Bowers, J. S., & Davis, C. J. (2012). Bayesian just-so stories in psychology and neuroscience. Psychological Bulletin, 138(3), 389-414. https://doi.org/10.1037/a0026450
Broadbent, D. E. (1958). Perception and communication. Pergamon Press. https://doi.org/10.1016/C2013-0-08164-9
Chomsky, N. (1959). [Review of the book Verbal behavior, by B. F. Skinner]. Language, 35(1), 26-58. https://doi.org/10.2307/411334
Clark, A. (2013). Whatever next? Predictive brains, situated agents, and the future of cognitive science. Behavioral and Brain Sciences, 36(3), 181-204. https://doi.org/10.1017/S0140525X12000477
Cowan, N. (2001). The magical number 4 in short-term memory: A reconsideration of mental storage capacity. Behavioral and Brain Sciences, 24(1), 87-114. https://doi.org/10.1017/S0140525X01003922
Craik, F. I. M. (2002). Levels of processing: Past, present... and future? Memory, 10(5-6), 305-318. https://doi.org/10.1080/09658210244000135
Craik, F. I. M., & Lockhart, R. S. (1972). Levels of processing: A framework for memory research. Journal of Verbal Learning and Verbal Behavior, 11(6), 671-684. https://doi.org/10.1016/S0022-5371(72)80001-X
Firestone, C., & Scholl, B. J. (2016). Cognition does not affect perception: Evaluating the evidence for top-down effects. Behavioral and Brain Sciences, 39, e229. https://doi.org/10.1017/S0140525X15000965
Gigerenzer, G., & Gaissmaier, W. (2011). Heuristic decision making. Annual Review of Psychology, 62, 451-482. https://doi.org/10.1146/annurev-psych-120709-145346
Hasson, U., Nastase, S. A., & Goldstein, A. (2020). Direct fit to nature: An evolutionary perspective on biological and artificial neural networks. Neuron, 105(3), 416-434. https://doi.org/10.1016/j.neuron.2019.12.002
Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in the world? Behavioral and Brain Sciences, 33(2-3), 61-83. https://doi.org/10.1017/S0140525X0999152X
Lake, B. M., Ullman, T. D., Tenenbaum, J. B., & Gershman, S. J. (2017). Building machines that learn and think like people. Behavioral and Brain Sciences, 40, e253. https://doi.org/10.1017/S0140525X16001837
Lavie, N., Hirst, A., de Fockert, J. W., & Viding, E. (2004). Load theory of selective attention and cognitive control. Journal of Experimental Psychology: General, 133(3), 339-354. https://doi.org/10.1037/0096-3445.133.3.339
Lupyan, G. (2015). Cognitive penetrability of perception in the age of prediction: Predictive systems are penetrable systems. Review of Philosophy and Psychology, 6(4), 547-569. https://doi.org/10.1007/s13164-015-0253-4
Mahon, B. Z., & Caramazza, A. (2008). A critical look at the embodied cognition hypothesis and a new proposal for grounding conceptual content. Journal of Physiology-Paris, 102(1-3), 59-70. https://doi.org/10.1016/j.jphysparis.2008.03.004
Mandler, G. (2002). Origins of the cognitive (r)evolution. Journal of the History of the Behavioral Sciences, 38(4), 339-353. https://doi.org/10.1002/jhbs.10066
Marr, D. (2010). Vision: A computational investigation into the human representation and processing of visual information. MIT Press. (Original work published 1982) https://doi.org/10.7551/mitpress/9780262514620.001.0001
Ma, W. J., Husain, M., & Bays, P. M. (2014). Changing concepts of working memory. Nature Neuroscience, 17(3), 347-356. https://doi.org/10.1038/nn.3655
McClelland, J. L., & Rumelhart, D. E. (1981). An interactive activation model of context effects in letter perception: I. An account of basic findings. Psychological Review, 88(5), 375-407. https://doi.org/10.1037/0033-295X.88.5.375
McClelland, J. L., Mirman, D., Bolger, D. J., & Khaitan, P. (2014). Interactive activation and mutual constraint satisfaction in perception and cognition. Cognitive Science, 38(6), 1139-1189. https://doi.org/10.1111/cogs.12146
Miller, G. A. (1956). The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review, 63(2), 81-97. https://doi.org/10.1037/h0043158
Miller, G. A. (2003). The cognitive revolution: A historical perspective. Trends in Cognitive Sciences, 7(3), 141-144. https://doi.org/10.1016/S1364-6613(03)00029-9
Neisser, U. (2014). Cognitive psychology (Classic ed.). Psychology Press. (Original work published 1967) https://doi.org/10.4324/9781315736174
Newell, A., & Simon, H. A. (1972). Human problem solving. Prentice-Hall.
Nosek, B. A., Hardwicke, T. E., Moshontz, H., Allard, A., Corker, K. S., Dreber, A., Fidler, F., Hilgard, J., Kline Struhl, M., Nuijten, M. B., Rohrer, J. M., Romero, F., Scheel, A. M., Scherer, L. D., Schönbrodt, F. D., & Vazire, S. (2022). Replicability, robustness, and reproducibility in psychological science. Annual Review of Psychology, 73, 719-748. https://doi.org/10.1146/annurev-psych-020821-114157
Oberauer, K., Lewandowsky, S., Awh, E., Brown, G. D. A., Conway, A., Cowan, N., Donkin, C., Farrell, S., Hitch, G. J., Hurlstone, M. J., Ma, W. J., Morey, C. C., Nee, D. E., Schweppe, J., Vergauwe, E., & Ward, G. (2018). Benchmarks for models of short-term and working memory. Psychological Bulletin, 144(9), 885-958. https://doi.org/10.1037/bul0000153
Open Science Collaboration. (2015). Estimating the reproducibility of psychological science. Science, 349(6251), aac4716. https://doi.org/10.1126/science.aac4716
Posner, M. I. (1980). Orienting of attention. Quarterly Journal of Experimental Psychology, 32(1), 3-25. https://doi.org/10.1080/00335558008248231
Ritter, F. E., Tehranchi, F., & Oury, J. D. (2019). ACT-R: A cognitive architecture for modeling cognition. Wiley Interdisciplinary Reviews: Cognitive Science, 10(3), e1488. https://doi.org/10.1002/wcs.1488
Sperling, G. (1960). The information available in brief visual presentations. Psychological Monographs: General and Applied, 74(11), 1-29. https://doi.org/10.1037/h0093759
Treisman, A. M., & Gelade, G. (1980). A feature-integration theory of attention. Cognitive Psychology, 12(1), 97-136. https://doi.org/10.1016/0010-0285(80)90005-5
Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases. Science, 185(4157), 1124-1131. https://doi.org/10.1126/science.185.4157.1124
Wilson, M. (2002). Six views of embodied cognition. Psychonomic Bulletin & Review, 9(4), 625-636. https://doi.org/10.3758/BF03196322
Wilson, R. C., & Collins, A. G. E. (2019). Ten simple rules for the computational modeling of behavioral data. eLife, 8, e49547. https://doi.org/10.7554/eLife.49547
Wolfe, J. M. (1994). Guided Search 2.0: A revised model of visual search. Psychonomic Bulletin & Review, 1(2), 202-238. https://doi.org/10.3758/BF03200774