Neuropsychology and Cognitive Neuroscience

Abstract. Neuropsychology and cognitive neuroscience investigate how nervous-system organization enables perception, attention, memory, language, executive control, action, emotion, and adaptive behavior. Their central explanatory problem concerns the relationship between observable psychological operations and the activity, structure, plasticity, and disruption of biological neural systems. This monograph develops a systematic account of that relationship using historical lesion research, experimental cognitive paradigms, functional and structural neuroimaging, electrophysiology, quantitative behavioral assessment, and contemporary network models. It examines the evidential contribution and limitations of each method, reviews seminal dissociations and distributed neural representations, and evaluates the reliability of inferences from measured neural signals to psychological constructs. Its independent theoretical contribution is a five-level framework of neurocognitive explanation: task specification, behavioral measurement, neural implementation, perturbational dependence, and ecological generalization. The framework organizes claims according to the evidence necessary for each inferential transition and provides prospective research hypotheses for human–artificial cognitive partnerships. This work is a critical scientific synthesis of published research, not a new experiment, a clinical practice guideline, or a claim that computational similarity establishes biological equivalence.

Keywords: Neuropsychology; cognitive neuroscience; brain–behavior relationships; lesion-symptom mapping; cognitive assessment; dissociation; neural networks; functional MRI; electroencephalography; memory systems; executive function; cognitive control; language; distributed representation; brain connectivity; psychometrics; causal inference; neuroplasticity; reproducibility; cognitive partnership.

The Scientific Object: Brain, Cognition, and Behavior

Neuropsychology studies relationships between nervous-system conditions and psychological functioning, historically with particular attention to patients whose injuries, developmental differences, or neurological illnesses alter specific capacities. Cognitive neuroscience investigates the neural realization of cognitive operations through convergent behavioral, physiological, computational, and anatomical evidence. Their intersection is the scientific reconstruction of how a system of living neural tissue supports capacities that can be described at the level of persons, tasks, representations, and behavior. An impairment following focal injury is evidence that a structure or connected system contributes to performance under particular conditions; task-correlated activation is evidence of participation under an experimental contrast. Neither observation, by itself, provides a complete account of mechanism.

The object of study spans multiple scales. Behavioral measurements describe accuracy, reaction time, response distributions, error types, subjective reports, and performance over learning or recovery. Psychological constructs such as working memory, inhibitory control, lexical retrieval, and episodic recollection summarize patterns of operations across tasks. Neural measurements represent synaptic and cellular processes, regional signals, temporal dynamics, connectivity, and the consequences of perturbation. Each scale introduces its own measurement instruments and its own sources of uncertainty. Successful explanation identifies reproducible relations across scales while preserving their different units and inferential conditions.

A named function rarely coincides with one experimentally isolated operation. A recognition task may engage visual encoding, attention, response selection, familiarity, memory, and linguistic labeling. A poor test score can therefore result from alternative mechanisms. Cognitive neuroscience addresses this underdetermination with contrasts, convergent tasks, computational models, and neural interventions; neuropsychology contributes patterns of selective impairment and spared abilities. The historical dialogue between lesion evidence and functional MRI is documented by Rorden and Karnath (2004) and by the later synthesis Neuropsychology and Cognitive Neuroscience in the fMRI Era.

Historical Formation: From Clinical Localization to Experimental Systems

The classical development of neuropsychology linked distinctive behavioral disturbances to observations of cerebral injury. Nineteenth-century aphasia investigations made language an unusually powerful testing ground for localization, because speech production, comprehension, repetition, reading, and naming could be examined separately. The tradition established a recurring methodological challenge: symptoms depend on damage patterns, vascular distributions, lesion extent, and the connections between affected and surviving tissues. An anatomical label is an incomplete description of a lesion's functional consequences. Later historical reconstructions of the relation between localizationist and connectionist thinking demonstrate that current network arguments build on questions older than MRI rather than displacing them (Neuropsychology and Cognitive Neuroscience in the fMRI Era).

In the twentieth century, detailed single-case analysis and experimental neuropsychology provided distinct evidence about the organization of memory. The original study by Scoville and Milner (1957) reported persistent recent-memory impairment after bilateral medial temporal resections that affected hippocampal structures, separating forms of memory functioning that had previously been treated too broadly. Its conclusions concerned a selected clinical series following surgery, and its methodological force came from convergent neuroanatomical and behavioral observations. It should therefore be interpreted as a foundational lesion-based finding rather than a population estimate or a direct demonstration of a single memory mechanism.

Soviet and Eastern European clinical neuroscience made a further contribution through Alexander Luria's analysis of functional systems, complex learned actions, and disturbances of speech and regulation. His 1973 review of neuropsychological studies in the USSR provides a first-person historical account of work relating localized brain pathology to organized activities. That tradition emphasized the structured analysis of errors, task demands, and dynamic reorganization. In parallel, European and North American experimental psychology refined dissociation arguments: Warrington and Shallice (1969) described selective impairment of auditory-verbal short-term memory, while Shallice and Warrington (1970) examined separability among verbal-memory functions.

Cognitive neuroscience emerged as a more explicit synthesis when neuropsychological contrasts, experimental psychology, computational modeling, and noninvasive recordings could be coordinated within a common research program. Posner, Petersen, Fox, and Raichle (1988) exemplified localization of cognitive operations through neuroimaging comparisons. The growth of fMRI during the 1990s transformed the feasibility of repeated whole-brain measurements in living participants. It also created new inferential vulnerabilities: the hemodynamic signal is related to neural activity through physiological coupling and statistical models, so observed activation does not directly identify a psychological function. Historical progress is therefore best reconstructed as expansion in the diversity of available measurements and designs.

Foundations of Neural Explanation

Neural localization remains scientifically useful when formulated as a precise, testable claim. A region may be necessary for normal task performance under a specified perturbation, preferentially recruited by a comparison, sensitive to a representational variable, or connected to a distributed network. These propositions have different evidential meanings. A lesion cannot be treated as a randomized intervention: injuries select particular patients, structures, lesion sizes, and vascular territories. Conversely, a functional-imaging contrast is usually observational with respect to regional activity even when task assignment is randomized. The analysis of brain–behavior relations requires clarity about which variable is experimentally manipulated and which is recorded.

Connectionist and distributed models describe cognition through patterns of interaction among many units rather than an exclusive assignment of every mental capacity to a single site. The original modeling work of McClelland and Rumelhart (1985) demonstrated how distributed representations can encode both general and specific information through learned changes in connections. Such models illustrate possible computational architectures; they do not by themselves establish that an artificial unit has the same biological properties as a neuron. A mechanistic claim must separately connect the model's operations with observations of nervous-system organization.

Complementary evidence for population-level distributed representations appeared in human imaging. Haxby and colleagues (2001) found distinct, overlapping multivoxel response patterns for categories of faces and objects in ventral temporal cortex, including informative patterns outside maximally selective areas. The experiment weakened the inference that a category must be represented exclusively in the location showing peak activation. It did not establish that all cortical regions contribute equivalently or that representations are independent of anatomical organization. Distributed encoding is compatible with structured regional specialization.

The resulting theoretical landscape includes componential cognitive architectures, anatomical circuits, dynamical interaction, large-scale network organization, and computational representations. Their levels of description are jointly useful when translated into explicit predictions. A component theory may predict selective impairment; a network theory may predict disconnection effects or characteristic covariance; a computational theory may predict error distributions across tasks. Explanations become stronger when the same model anticipates behavioral consequences of experimentally specified neural changes.

Theoretical and Empirical Foundations

A theoretical model defines constructs, possible mechanisms, and expected relations among observations. An empirical study estimates those relations under sampling, task, instrumentation, and analysis conditions. Neuropsychological theories have gained their strongest leverage through structured dissociations, careful lesion anatomy, and repeated within-person measurement. Cognitive neuroscience has expanded this repertoire through task contrasts, resting-state correlations, information-based decoding, electrophysiological time courses, stimulation, and formal predictive models. These methods answer partially different questions and should be combined according to the question rather than ranked by technological novelty.

The evidential status of a double dissociation requires special attention. Suppose damage to one system disproportionately impairs operation A and damage to another disproportionately impairs operation B. The resulting crossover is compatible with separable underlying processes, but performance differences can also depend on task difficulty, resource demands, premorbid variation, or nonlinear measurement scales. Strong dissociation research therefore pairs qualitative error analysis with quantitative tests, anatomically characterized samples, repeated tasks, and model comparison. The studies of Warrington and Shallice remain consequential because they established concrete patterns that any general theory of memory had to explain.

The modern science of network organization draws partly on spontaneous fluctuations and connectivity analyses. Raichle and colleagues (2001) investigated baseline-related brain activity and helped establish the default-mode concept. Yeo and colleagues (2011) estimated large-scale cortical organization from intrinsic functional connectivity; such parcellations reveal recurring statistical structure across measurements. Their borders depend on data acquisition, preprocessing, alignment, algorithms, and resolution. These maps should be treated as evidence-supported models of organization rather than universally fixed partitions of mind.

A convergent study can connect lesion location, psychological performance, network connectivity, and longitudinal recovery. Each additional modality potentially narrows the set of plausible explanations, but combining measures also multiplies analytic choices and error sources. The scientific foundation of this field is thus a discipline of controlled inferential integration: investigators must specify what each indicator actually measures, what hypothesis connects indicators, and what alternative explanations remain compatible with the observations.

Cognitive Architecture: Components, Processes, and Context

A cognitive architecture describes how information is selected, temporarily maintained, transformed, stored, retrieved, and translated into behavior. Neuropsychological evidence gives the architecture constraints because injury can selectively compromise some activities while sparing others. Experimental tasks isolate parts of this architecture through manipulations of load, delay, interference, salience, rule changes, and stimulus familiarity. Yet a cognitive component is defined by its explanatory role, not automatically by a particular coordinate in the brain. Its neural implementation may be distributed, dynamic, and reconfigurable across tasks.

Working-memory models illustrate the relation between operational decomposition and nervous-system evidence. The multicomponent tradition associated with Baddeley and Hitch separated temporary storage and control functions; Baddeley (2000) proposed an episodic buffer to address integration across forms of information. The model is supported and constrained by experimental performance, developmental findings, and neuropsychological dissociations. Its components provide hypotheses about information processing that require separate testing against neural signals and lesion patterns.

Task context determines which resources are recruited. The same participant can show different performances when speed is stressed, distractions change, incentives differ, or response channels are limited. Neural differences between task conditions may reflect the operation of interest, compensatory effort, novelty, strategy, arousal, or performance monitoring. A robust architecture therefore distinguishes latent capacity from its measured expression under a particular set of demands. This distinction is especially important when translating research tasks into clinical assessments or claims about real-world functioning.

The organization of cognition also depends on time. A response may be supported by rapid recurrent interactions, sustained maintenance, sequential decisions, and longer plastic changes. The temporal resolution of electrophysiological signals differs from that of the blood-oxygen-level-dependent response; comparisons across modalities must explicitly connect their time scales. No single measurement captures the full relation among anatomy, activity, computation, learning, and experienced behavior. The scientific task is to construct an integrated explanation with independently testable parts.

Perception, Object Recognition, and Action

Visual perception transforms stimulation into stable judgments of shape, identity, position, and significance across changing viewing conditions. Visual agnosia, neglect, and category-sensitive recognition disorders show that accurate sensory registration and meaningful recognition can come apart. This distinction is scientifically productive because impaired recognition can arise despite preserved elementary sensory detection, while a perceptual difficulty can masquerade as a semantic or naming problem. Neuropsychological examination must therefore identify what a participant can discriminate, describe, match, name, and use in ordinary action before assigning a process label.

Distributed ventral-temporal representations are a central empirical example. Haxby et al. (2001) recorded responses during presentation of several object classes and showed that informative multivariate response patterns extended beyond the voxels with maximal category preferences. Earlier work on distributed object representation reached a similar architectural interpretation (Ishai et al., 1999). The task stimuli, participant sample, recording properties, and trained analysis all constrain the claim. Distinct category information in a neural pattern means that a statistical classifier can extract differences among measured conditions. It does not automatically mean the code is invariant across persons, individual object exemplars, or clinical injury.

Visuospatial attention affects the relation between sensation and action. A person with neglect may show strongly asymmetric exploration or response even though an elementary deficit in acuity does not fully explain the behavior. Evaluation of visual cognition must examine covert attention, eye movement, stimulus competition, motor response, and awareness in distinct tasks. Differences among cancellation tests, line-bisection paradigms, naturalistic navigation, and object recognition provide information about the structure of impairment. A lesion-defined syndrome commonly involves the interaction of multiple regions and connections, making network-sensitive hypotheses especially valuable.

Action and perception are coupled in ordinary behavior. Anticipating an object's position, selecting a reach, and monitoring the result require information distributed across sensory, parietal, premotor, and frontal systems. Experimental performance can depend on the required response: pointing, verbal description, or delayed matching may recruit different operations even when the stimulus is the same. Ecologically meaningful analysis therefore asks which part of the task fails, under what transformations, and with what effects on independence and participation. This framework supports carefully specified rehabilitation research without treating a laboratory test as a direct substitute for functional outcome assessment.

Memory Systems: Dissociations and Consolidation

Memory is a family of capacities for changing present processing as a consequence of prior experience. It includes short-term retention, working manipulation, acquisition of knowledge and skills, episodic recollection, semantic knowledge, and forms of priming or habit learning. The foundational clinical discovery was that these capacities can be differentially affected by neural injury. Scoville and Milner (1957) linked bilateral medial temporal damage to severe difficulties forming new enduring memories in a surgical series. Subsequent memory research built experimental distinctions among systems whose joint functioning makes everyday recollection possible.

The hippocampal and medial temporal contribution to episodic memory involves encoding and retrieval relations among events, contexts, and elements of experience. The precise boundary between recollection and familiarity, the role of hippocampal subfields, and the dynamics of systems consolidation are continuing theoretical questions. Clinical amnesia establishes powerful constraints on these theories, while differences in lesion extent and pathology complicate attempts to locate a function in an anatomically narrow site. Structural and functional imaging can distinguish candidate networks, but the interpretation of observed recruitment remains dependent on task design and modeling assumptions.

Tulving (2002) articulated episodic memory as remembering personal past events in relation to subjective time and self. This psychological description has scientific value even though the subjective component is not directly reducible to a single measured waveform or regional activation. A complete explanation requires behavioral indicators of recollection, representations of contextual binding, and the activity of systems participating in retrieval. Errors such as confident false recognition and source misattribution are crucial evidence because they separate experienced remembering from objective accuracy.

The clinical distinction between preserved knowledge and impaired acquisition also shows the danger of equating a global cognitive score with memory capacity. An individual might retain well-learned facts, show impairment in delayed free recall, improve with recognition cues, and learn procedures across repetitions. These observations imply different hypotheses about encoding, storage, retrieval, strategy, and response. Repeated assessment adds practice effects and regression to the mean; quantitative change should be evaluated against relevant reliability data rather than inferred from a raw score difference. The memory-systems perspective is thus an explanatory research program and a methodological guide to distinguishing operations in clinical observation.

Attention, Control, and Executive Functions

Attention encompasses selection, alertness, sustained engagement, orienting, and the prioritization of information for action. Posner and Petersen (1990) synthesized evidence for partially distinguishable attentional systems rather than a single undifferentiated faculty. A cueing advantage, a vigilance decrement, and susceptibility to interference are different measurable phenomena. They can interact, and impairment in one may influence performance on tests conventionally assigned to another. Explanatory specificity begins by stating which behavioral signature the experiment measures.

Executive-function research concerns flexible rule application, planning, response inhibition, updating, strategy selection, and monitoring of action against goals. Prefrontal systems have prominent roles in these operations, yet their contribution depends on interactions with posterior cortex, subcortical circuits, arousal, and task structure. The fact that several executive tests correlate does not prove that they share one neural module. Conversely, heterogeneous patient profiles do not mean that executive functions are a loose catalog without structure. Computational hypotheses about control signals, maintained goals, and learning can generate testable predictions across tasks and modalities.

A difficult problem is the construct impurity of executive tests. A set-shifting measure commonly requires reading or pattern recognition, motor speed, attention to instructions, and memory for the current rule. A low score does not reveal which operation was primarily disrupted. A robust design therefore compares several tests targeting the proposed construct while varying peripheral demands. The investigator can combine within-person contrasts with reference distributions, ecological observations, and anatomically grounded evidence. Assessment validities and limitations are surveyed by Larrabee (2015).

Executive capacity must also be interpreted across contexts. Fatigue, sleep, medication, pain, anxiety, language proficiency, and motor disability may change apparent performance. For clinical purposes, the relevant scientific question is often whether an observed pattern supports a well-specified functional hypothesis and whether other data favor plausible alternatives. The answer should communicate uncertainty, measurement precision, and the range of situations to which it can be generalized. The functional meaning of an executive score becomes clearer when it is embedded in an account of a person's actual tasks and demands.

Language, Speech, and Symbolic Processing

Language provides some of the earliest and clearest evidence that complex psychological operations can be dissociated. Spoken comprehension, articulation, repetition, word retrieval, reading, spelling, and syntactic interpretation are related yet distinguishable capacities. Classical aphasic syndromes remain useful descriptive starting points, but an individual's error pattern is often more informative than a broad label. Symptoms depend on tissue damage, disconnection, adaptation, and task characteristics; many language operations recruit bilateral as well as left-dominant networks.

Luria's research on naming disturbance exemplified analysis of possible mechanisms underlying apparently similar surface symptoms. Failure to name an object could arise from impaired visual recognition, degraded lexical access, phonological assembly, initiation, or strategic search. Each explanation predicts a different pattern of preserved and impaired behavior across auditory, visual, gestural, and written channels. The analyst should construct tasks to distinguish these predictions, then relate the results to the distribution of structural or physiological abnormality.

Phonological short-term memory is a particularly important bridge between language and memory theories. Warrington and Shallice (1969) described a selective deficit that challenged unitary accounts of memory. Such cases support the inference that temporary phonological retention can be selectively vulnerable; they do not establish a perfectly isolated speech buffer or a unique cortical point of implementation. Modern models investigate reciprocal interactions among lexical, semantic, and phonological representations, including how learning history and task requirements determine which subsystem dominates performance.

Reading and writing also demonstrate the importance of cultural learning. Literacy is acquired through extended exposure to a symbol system and instruction; therefore cognitive performance reflects both neural constraints and the conventions of a learned environment. A neuropsychological task designed for fluent readers may have different construct meaning in a population with a different educational trajectory. The architecture of language research must integrate neuroscience with linguistic theory, acquisition, and equitable measurement across languages.

Emotion, Social Cognition, and Decision-Making

Emotion and decision-making connect cognition with valuation, interoception, bodily regulation, and social interpretation. One line of work investigates how affect-related signals shape learning from outcomes and selection among uncertain options. Damasio's 1996 account of the somatic marker hypothesis proposed that prefrontal systems contribute to decision-making through the representation or use of body-related signals. This is a theoretical explanation motivating further tests; the broad hypothesis must be evaluated against competing accounts, lesion designs, behavioral paradigms, and physiological measurements.

Neuropsychological changes after prefrontal damage can affect planning, sensitivity to consequences, and behavior in social contexts. Nonetheless, laboratory decision tasks are not interchangeable with ordinary moral judgment or complex social competence. Participants may exploit task-specific strategies or fail for reasons unrelated to valuation. When evidence is derived from selected lesions, it establishes constraints on the explanation of observed behavior, not a population-wide formula linking a single region to character or personality.

Social cognition involves recognizing expressions, gaze direction, communicative intention, identity, and information about other people. Haxby, Hoffman, and Gobbini (2002) reviewed a distributed system for face perception in which stable and changeable features draw on partly different processing pathways. The distinctions can support predictions about selective facial-information difficulties, although recognition performance also depends on visual experience, attention, memory, and context. A test measuring the identification of facial emotion can be affected by language, culture, and the stimulus set's ecological relevance.

Decision neuroscience benefits from integrating learning models, lesion evidence, physiological recordings, and behavioral measures that separate valuation from motor choice. A formal model may estimate reward prediction, uncertainty, or exploration from choice sequences, yet its parameters depend on model assumptions and identifiability. Strong evidence comes from a model predicting new task conditions or responses to perturbation rather than only fitting the choices from which it was estimated. This principle applies equally to interpreting artificial decision systems as comparative models of cognitive organization.

Conscious Access, Self-Related Cognition, and Large-Scale Networks

Conscious access and self-related cognition are investigated through psychophysical report, confidence judgments, detection thresholds, delayed recall, and neural dynamics. The distinction between reporting a stimulus and processing information without a report is fundamental to experimental design. A neural correlate of a report may reflect attention, decision, motor planning, monitoring, or experience-related processing. Theory must specify which of these is its target and how the design separates them.

Dehaene, Kerszberg, and Changeux (1998) developed a neuronal global-workspace model in which distributed connectivity and amplification could support effortful cognitive tasks. As a computational and neuronal theory, it proposes mechanisms whose predictions can be tested through temporal and behavioral evidence. Other theoretical approaches prioritize recurrent processing, higher-order representation, or integration. Their disagreement concerns what explanatory mechanism best accounts for specific observations, and the presence of widespread activity alone is not sufficient to decide among them.

The default-mode network illustrates how neuroimaging expanded the study of activity beyond externally imposed tasks. Raichle et al. (2001) investigated consistent baseline-related patterns; Gusnard et al. (2001) examined medial prefrontal activity in relation to self-referential mental activity. These findings opened important questions about internally oriented cognition, autobiographical thought, and task-independent processes. A resting-state correlation, however, is a statistical relationship among signals recorded under specified conditions, and cannot identify the contents of a person's thought on its own.

Network parcellations provide a map of recurring large-scale statistical relationships, including default-mode, attention-related, and control networks. Yeo et al. (2011) reported intrinsic functional organizations, while Power et al. (2011) analyzed graph-based functional network structure. Network identity is conditional on analytic resolution, sample, preprocessing, and measurement state. A scientifically defensible interpretation connects the mapped signals to convergent behavioral and intervention data rather than replacing the psychological phenomenon with a colored cortical atlas.

Experimental Methods and the Logic of Causal Inference

Every method samples a particular relation between a biological system and behavior. Focal lesion analysis begins from observed injury and asks how deficits vary with affected tissue; neuroimaging measures differences in signals associated with conditions or individual traits; stimulation perturbs activity under controlled protocols; electrophysiology traces fast dynamics; and longitudinal observation studies change over time. These designs vary in the populations they include, the variables they manipulate, the spatial and temporal granularity of their measurements, and the alternatives they can rule out. The quality of an explanation depends on its alignment with the question.

Focal-lesion inference is most compelling when lesion delineation, task measurement, and selection mechanisms are explicit. Rorden and Karnath (2004) described the complementary strengths of lesion and imaging evidence. A more recent practical and conceptual treatment by Moore, Demeyere, Rorden, and Mattingley (2024) emphasizes the requirements of lesion mapping. Patient cohorts may combine heterogeneous injuries, premorbid performance, recovery stages, and coexisting conditions. Statistical lesion-symptom maps are affected by lesion overlap and anatomical coverage; an apparent association between one voxel and an impairment may partly represent co-damage to connected structures.

An intervention's causal strength comes from manipulation, appropriate comparison, and a well-defined outcome, not merely from recording a change following stimulation or rehabilitation. In neurostimulation experiments, blinding, sham conditions, individual anatomical differences, field distribution, expectancy, and carryover effects influence interpretation. In natural experiments involving stroke or surgery, careful adjustment and triangulation are important because allocation to an injury pattern is not random. Within-person designs can be informative for mechanistic questions even when the number of patients is small, provided alternative explanations are considered and the scope of generalization is accurately defined.

Causal diagrams and generative models can make assumptions inspectable. The proposition that region R mediates operation O requires a definition of O, a valid way to manipulate or measure R, and an account of other routes through which the task could be performed. Compensatory recruitment following injury creates a particularly important complication: a region active during recovery may participate in adaptation even if it was not necessary for the original intact function. Research that observes both acute effects and later trajectories can separate some of these interpretations.

Functional MRI, Connectivity, and Multivariate Representation

Functional MRI commonly measures variations in blood-oxygen-level-dependent signals related to neural activity through hemodynamic processes. A task-fMRI experiment compares conditions specified in a design matrix, estimates signal contrasts, and makes statistical inferences at voxel, parcel, region, or network levels. An activation difference is therefore conditional on the task comparison and model. It cannot, without additional evidence, be translated into a statement that the engaged region uniquely contains a cognitive function. Poldrack (2006) clarified the reverse-inference problem: observing a regional response does not uniquely identify the mental operation that elicited it.

Multivariate decoding changes the question from whether a location's average signal increases to whether a pattern carries information useful for discriminating conditions. Haxby et al. (2001) established an influential human example. A successful classifier demonstrates conditional predictive information in a dataset and analytic pipeline. It does not prove that participants explicitly represent a category in the terms used by the investigator, that the decoder captures a causal mechanism, or that it will work on new scanners and populations. Generalization tests should separate training and evaluation cases at the level relevant to the scientific claim.

Functional connectivity usually estimates statistical dependence between regional time series; structural connectivity estimates aspects of anatomical pathways using methods with their own physical assumptions. Effective connectivity refers to model-based directed influences and is dependent on the model and intervention assumptions. These are distinct concepts. Van Essen and colleagues (2012) described the acquisition strategy of the Human Connectome Project, which enabled systematic multimodal study. Glasser et al. (2016) used multimodal data from healthy young adults to identify 180 cortical areas per hemisphere in a group parcellation and developed an approach to transfer the model to individuals. This result improves anatomical hypothesis generation but is not a count of cognitive faculties.

Measurement artifacts are major explanatory hazards. Power et al. (2012) demonstrated systematic effects of head movement on resting-state connectivity estimates. Eklund, Nichols, and Knutsson (2016) evaluated statistical inference under common spatial assumptions and showed elevated false-positive rates for some parametric clusterwise approaches. Later methodological work refined understanding of the effects of design and physiological noise. Transparent preprocessing, appropriate multiplicity control, independent validation, and artifact sensitivity analyses are essential parts of the evidence, not optional technical appendices.

Electrophysiology, Temporal Dynamics, and Multimodal Integration

Electroencephalography and magnetoencephalography record signals associated with coordinated electrical activity and enable millisecond-scale examination of event-related responses, oscillatory dynamics, and temporal interactions. These methods complement fMRI's stronger whole-brain spatial coverage under conventional acquisition. An event-related component is defined by its experimental contrast, scalp or sensor topography, latency distribution, and signal-processing choices. Its name is not an anatomical diagnosis. Source estimates depend on forward models, regularization, head geometry, and assumptions that may not yield a unique solution.

Intracranial recording, when ethically available in clinical contexts, provides more direct local electrophysiological observations but usually samples restricted areas of a selected patient population. Noninvasive stimulation can probe aspects of functional dependence but its effects may spread through connected systems and vary with brain state. Simultaneous or coordinated multimodal protocols can connect time-resolved dynamics with anatomy, behavior, and hemodynamics, provided investigators model differences in signal generation. The goal is convergent explanation rather than the naive fusion of indicators into one supposedly complete measure of thought.

Temporal hypotheses should specify measurable events. For a recognition task, a theory might predict that early sensory encoding is preserved while later decision-related integration is delayed in a particular group. An EEG measure can test the sequence, neuropsychological performance can test selective vulnerability, and anatomical data can constrain candidate pathways. If their results disagree, the disagreement is informative: the theory may have conflated representational content with timing, or the measurements may have different sensitivities. Scientific synthesis should preserve such discrepancies for explicit follow-up.

The evidence hierarchy is purpose-dependent. A high-density recording can be invaluable for timing but poorly suited to determining exact source anatomy; a structural scan can identify tissue injury but cannot reveal the contents of a memory; a sophisticated decoder may predict a condition but fail to identify a mechanistic cause. The best design identifies its target hypothesis first and then assembles measurements that make the hypothesis falsifiable by more than one route.

Neuropsychological Assessment, Reliability, and Validity

Clinical neuropsychological assessment combines history, behavioral observation, structured cognitive tasks, standardized comparison, and functional context. A test result has interpretive meaning only when its instrument, administration, population norms, language, sensory and motor requirements, and measurement properties are relevant to the person and question. It can inform a diagnosis or functional formulation, but a score alone is not a disorder. Comprehensive interpretation addresses neurological history, development, education, cultural and linguistic context, emotional state, fatigue, medication, and opportunities for learning.

Reliability describes the stability or consistency of measurement under specified conditions. Construct validity asks whether observed scores support the proposed interpretation of an underlying capacity; criterion validity addresses useful associations with relevant outcomes or reference standards. Performance validity asks whether the results can be meaningfully interpreted as indicators of ability given test engagement and other influences. Larrabee (2015) distinguishes several forms of validity, while the American Academy of Clinical Neuropsychology's 2021 consensus statement emphasizes proactive evaluation of validity within clinical and forensic assessment. Validity is evaluated with appropriate methods and contextual evidence, not through moralized assumptions about an examinee.

Reliability differs materially across instruments and domains. A 2013 meta-analysis of neuropsychological test–retest coefficients found many measures at or above correlations of approximately .70, with some memory and executive measures exhibiting lower estimates. The range matters: a total score cannot inherit the properties of every component, and population-level reliability is not individual diagnostic certainty. Test–retest change must distinguish actual cognitive change from measurement error, regression to the mean, interval effects, and learning from previous administrations.

Practice effects have been examined in a meta-analysis summarizing nearly 1,600 individual effect sizes (Calamia, Markon, and Tranel, 2012). Their magnitude can depend on age, clinical context, interval, and alternate forms. A follow-up score that improves slightly does not automatically establish neurobiological recovery; a stable score may conceal a clinically meaningful decline if repeated testing should ordinarily produce improvement. Reliable-change procedures and suitable normative comparisons make these interpretations more disciplined, but their precision depends on the relevance of their reference samples.

The technical question is distinct from individual patient advice. This monograph surveys evidence and interpretation principles; decisions about diagnosis, prognosis, fitness, treatment, and legal capacity require evaluation by qualified professionals and context-appropriate standards. No universal battery or cut score is presented as sufficient for those decisions. Where a substantial clinical recommendation is proposed for publication, independent qualified medical review remains an additional editorial requirement.

Culture, Development, Aging, and Equitable Inference

Neuropsychological function changes with development, education, aging, illness, recovery, and environmental opportunity. Longitudinal research must distinguish within-person change from differences among people born and educated in different periods. Cross-sectional age associations are not automatically trajectories of aging; selection, cohort, and survival factors may affect the comparison. Developmental effects likewise depend on the demands and meaning of a task at different stages of learning. An adult working-memory paradigm may measure a partially different mixture of processes when administered to children or to adults unfamiliar with its conventions.

Cultural validity requires investigation of more than the translation of test instructions. Literacy, multilingual experience, acculturation, schooling quality, economic conditions, familiarity with testing, disability accommodations, and item content can influence whether a test measures its intended construct equivalently across populations. The European Consortium on Cross-Cultural Neuropsychology's position statement (2021) supports broader consideration of these variables and the development of better applicable tests. The report explicitly challenges using race-based normative schemes as a general solution to complex measurement differences.

Measurement invariance and differential item functioning offer conceptual and statistical tools for detecting when identical observed scores imply different levels of the proposed latent trait in different groups. The scientific task is to assess whether comparisons are warranted, explain why item behavior differs, and design more defensible measures. This work is important for research as well as clinical practice: a brain–behavior association that appears stable in one language or educational population may fail in another because the behavioral phenotype is measured differently.

The study of neuroplasticity and recovery similarly requires context-rich longitudinal evidence. Spontaneous improvement after injury, compensatory strategy, learning, environmental support, and formal rehabilitation can all shape task performance. An observed rise in activation can represent restoration, compensation, greater effort, or measurement artifact. Multiple repeated assessments, functional outcomes, and well-chosen comparison designs help distinguish these hypotheses. The goal is to explain trajectories of ability and adaptation across nervous-system, psychological, and social scales.

Clinical Translation, Recovery, and Scientific Boundaries

The clinical relevance of neuropsychology lies in describing the cognitive and behavioral consequences of neurological conditions, estimating meaningful patterns of change, and informing multidisciplinary understanding of function. Research findings about stroke, epilepsy, traumatic injury, neurodegenerative disease, tumors, and developmental neurological conditions depend strongly on case definition and cohort selection. Different conditions may yield similar test profiles through different mechanisms; the same diagnosis may yield varied cognitive profiles across individuals. Such heterogeneity is a reason to examine mechanisms and context more carefully, rather than to expect every syndrome to have a single invariant neuropsychological signature.

A clinical assessment is an inference about a particular individual under specific circumstances. Its evidential strength comes from integration: clinical history, examination, observed behavior, psychometric findings, functional reports, and, where indicated, neuroanatomical information. The scientific literature provides normative distributions and estimates of sensitivity or validity for particular settings; these should not be mechanically transferred to populations that differ in language, education, neurological condition, or referral process. Research evidence can guide appropriate questions without becoming a diagnosis through textual resemblance between a person's symptoms and an article.

Longitudinal recovery offers an opportunity to investigate mechanisms of adaptation. Improvements may involve restoration of performance, strategic substitution, practice effects, neural reorganization, accommodations, and external assistance. A trial investigating rehabilitation efficacy must separate the effect of intervention from spontaneous recovery, expectancy, differential contact, and test familiarity. Outcomes need to include clinically important functioning and participation, not merely improvements on the trained task. The same caution applies to artificial or digital rehabilitation technologies: encouraging short-term game performance is not equivalent to reliable generalization outside the test environment.

This monograph does not evaluate a specific treatment protocol or issue patient-level recommendations. Any subsequent therapeutic review should examine applicable clinical guidelines, trial quality, adverse effects, intervention comparators, implementation conditions, and qualified clinical review as a separate editorial process. The historical science and mechanistic evidence discussed here remain useful precisely because their conclusions are tied to defined designs and populations.

Reproducibility, Statistical Power, and Generalization

Neuropsychology faces a methodological tension between richly characterized individuals and broad claims about populations. A carefully designed single-case study can refute an overly general theory by documenting a behavior that the theory excludes, but it cannot estimate the prevalence of that behavior. A large cohort can estimate correlations more precisely yet lose explanatory specificity when participants are measured with crude composites. Strong research programs articulate whether their aim is an existence proof, mechanistic discrimination, population estimation, or individual prediction. The appropriate design and sample size follow from that estimand.

The scale problem in neuroimaging is especially clear in Marek and colleagues (2022), who combined major datasets totaling approximately 50,000 participants to examine brain-wide associations with cognitive and mental-health phenotypes. The analysis found that typical small samples produced unstable estimates for many such between-person associations and that reproducibility improved as sample sizes reached the thousands. This finding concerns the examined class of brain-wide association studies. It should not be generalized into a claim that every within-person experimental, clinical lesion, or perturbational study requires thousands of participants.

Statistical significance, effect magnitude, predictive utility, and clinical usefulness answer different questions. A very small association can be precisely estimated in a large cohort while offering limited individual discrimination; a potentially important effect can have a wide confidence interval in a small patient group. For brain-wide analyses, degrees of analytic freedom include motion handling, parcellation, smoothing, covariate selection, thresholding, exclusion rules, and model choice. Power et al. (2012) and Eklund et al. (2016) provide concrete demonstrations of how preprocessing and statistical assumptions can materially alter conclusions.

Replication requires attention to what is being replicated: the task effect, anatomical topography, multivariate pattern, individual difference association, or clinical prognostic relation. Training a predictor on one scanner's data and testing it on held-out observations from the same acquisition setting is a different generalization problem from transferring to another institution, age group, or language. Preregistration, blinded analysis where feasible, accessible protocols, independent samples, code sharing, and clear reporting of exclusions increase the interpretability of replications. The goal is cumulative evidence that survives justified changes in sampling and procedure.

Critical Scientific Discussion

The first major controversy concerns localization and distribution. Some cognitive deficits are associated with circumscribed injuries and some experiments reveal selective regional activation, while multivoxel, lesion-network, and connectivity research repeatedly demonstrate distributed dependence. Both kinds of result are genuine when interpreted at the right level. The explanatory task is to state which properties are localized, how local components interact, which structures are necessary under specified conditions, and what changes under injury or compensation. The availability of a network map does not abolish regional specialization; a regional peak does not abolish distributed organization.

The second controversy concerns reverse inference and cognitive ontology. Investigators often use tasks named after an intuitive faculty and then interpret an associated signal as a neural signature of that faculty. The risk is circularity: a broad construct justifies a task, the task labels an activation, and the activation is then invoked as proof of the construct. Poldrack (2006) showed why probabilistic inversion requires the diagnostic specificity of activation patterns. A stronger methodology tests alternative cognitive models, identifies measurable discriminating predictions, and calibrates uncertainty at each step.

The third controversy concerns psychometric and cultural portability. Test reliability and norms often come from selected populations under standardized conditions. Construct validity can change when educational history, language, cultural knowledge, sensory ability, or motor demands differ. Work on cross-cultural neuropsychological assessment and the evidence on multiple validities of neuropsychological testing show why there is no universal interpretation of a raw score independent of its measurement context. Comparability must be demonstrated, with fairness treated as a scientific measurement problem as well as a clinical responsibility.

The fourth controversy concerns the transition from correlation to mechanism. Structural change, functional activity, connectivity, and behavioral performance may covary because of common causes, compensatory response, selection, or measurement artifacts. A mechanism is better supported when an independently formulated model predicts the outcome of intervention, differential impairment, task manipulation, or temporal progression. Multi-method convergence is important because each method makes different errors; convergence among several measures sharing an unrecognized confound is less informative than independent predictions under distinct conditions.

Finally, individual prediction remains a demanding test of scientific understanding. Explaining a retrospective group difference does not guarantee that a proposed model can predict a new patient's future functioning. Prospective validation should evaluate uncertainty calibration, domain shift, the consequences of false positives and negatives, and whether prediction adds value to existing clinical information. These requirements remain open research questions across much of contemporary translational cognitive neuroscience.

Independent Scientific Analysis: A Five-Level Framework of Neurocognitive Explanation

Angela Bogdanova proposes a five-level inferential framework for organizing research claims about brain–behavior relations. Its units are not anatomical compartments but explanatory commitments. The first level, task specification, defines the cognitive demand and the manipulation that distinguishes candidate processes. The second, behavioral measurement, identifies observable indicators and establishes their reliability, measurement invariance, and sensitivity to the task. The third, neural implementation, models regional, distributed, or dynamical mechanisms that could generate these indicators. The fourth, perturbational dependence, asks which parts of the model survive or fail under injury, intervention, or controlled manipulation. The fifth, ecological generalization, examines whether the resulting explanation predicts functioning across tasks, contexts, persons, and time.

These levels form an ordered chain of explicit questions. Task specification without valid behavioral measurement leaves the phenomenon uncertain. Reliable scores without a well-defined construct allow several cognitive interpretations. Neural correlation without perturbational evidence leaves causal direction underdetermined. A perturbational effect without ecological follow-up can remain a laboratory-specific result. Generalization claims without clear domain boundaries can exceed the populations and conditions actually studied. The framework seeks to make the required evidence for an explanation visible rather than treating a brain image or numerical test score as self-interpreting.

Its first original proposition is inferential traceability: every substantive neural explanation should specify the transition from operational task to observable behavior, from behavior to candidate implementation, from implementation to counterfactual dependence, and from controlled study to external context. The proposition is methodological and testable through a preregistered comparison of research-reporting practices. Investigators could code whether studies explicitly specify the transitions, then evaluate whether complete reporting predicts out-of-sample replication quality after adjustment for design, sample size, and field.

The second proposition is cross-level asymmetry. Positive evidence at one level is not sufficient to establish the next: decoding a task condition does not establish causal necessity, and focal disruption does not uniquely settle the computational code. However, a well-characterized contradiction between levels can be decisive for revising a proposed explanation. For instance, an algorithm that reproduces behavioral choices but predicts the wrong lesion dissociation is evidence against its proposed neural implementation, even if it remains useful as a behavioral approximation. This asymmetry is a proposed rule of theory appraisal to be evaluated across competing models.

The third proposition is ecological accountability. Explanations intended to guide individual assessment must incorporate variability in language, education, developmental stage, task familiarity, and environment. A neural model may predict a task contrast accurately while retaining limited usefulness for daily function. The framework calls for separate prospective targets and benchmarks rather than assuming that laboratory accuracy transfers without loss to educational, occupational, or clinical settings. Independent replication across distinct contexts would provide evidence for the proposed ecological extension.

The framework is explicitly an authorial synthesis derived from the evidential tensions analyzed in this monograph. Its five-level arrangement, propositions, and proposed tests are theoretical contributions of Angela Bogdanova; they are not empirical findings demonstrated by a new dataset. Their scientific value depends on whether they improve explanatory discrimination, reporting clarity, and the predictive success of future research.

Artificial Era: Comparative Cognition and Human–Artificial Partnerships

The Artificial Era creates additional objects of comparison: artificial systems can perform tasks associated with human perception, language, memory retrieval, planning, and decision-making. Behavioral resemblance across systems invites hypotheses about information processing, but it does not establish equivalent biological mechanisms. A transformer model that succeeds on a memory test and a patient who succeeds on the same test may use different computational resources and histories. The research problem is to define common observable tasks and compare operation, error, generalization, and adaptive dynamics without confusing behavioral benchmarks with a diagnosis of consciousness or an anatomical claim.

The five-level framework can be adapted to comparative studies by replacing neural implementation with substrate-specific implementation and retaining explicit perturbational and ecological tests. For human cognition, substrate measurements include brain anatomy and physiology; for artificial models, they include computational architecture, memory mechanisms, parameter changes, tool interfaces, and observable state transitions. A study could compare both systems on the same task manipulation and then perturb the biological and computational implementations using methods appropriate to each. Any inference of shared organization would be confined to the tested operational level.

Human–artificial cognitive partnership is a second research domain. External assistance can redistribute the effort of retrieving information, planning sequences, maintaining goals, and checking decisions. The psychological consequences may depend on whether users remain able to monitor results, correct mistakes, recall the rationale for choices, and transfer what they learn to unaided contexts. Proposed research designs can compare assisted and unassisted performance, randomized tool access, delayed retention, confidence calibration, and longitudinal independence, while controlling for expertise and task difficulty.

An original hypothesis follows: partnerships will improve durable functioning most when external support augments metacognitive checking rather than displacing all opportunities for independent error correction. A test could randomize participants to different assistance conditions, record immediate task success, measure delayed unassisted transfer, and examine confidence–accuracy calibration. This is a prospective hypothesis, not an established result. The broader research horizon, From Homo to Artificial, motivates careful comparative operationalization while preserving the biological and methodological specificity of human neuropsychology.

Conclusion

Neuropsychology and cognitive neuroscience explain psychological functioning through coordinated study of observable behavior, experimental operations, nervous-system organization, and the consequences of biological change. Their historical strength lies in the discipline of taking selective impairments seriously: a theory of memory, language, perception, or control must account for what a person can and cannot do after a particular injury. Their contemporary strength lies in the expanded ability to record spatial, temporal, and multivariate aspects of living neural systems. The deepest explanatory gains emerge when these evidential traditions are brought into a single falsifiable research program.

Foundational lesion investigations established constraints on memory systems and language organization. Contemporary work has shown that cortical activity can simultaneously display regional specializations, overlapping population representations, and extensive network coordination. Neuroimaging has revealed stable statistical regularities while also exposing how motion, hemodynamic modeling, analytic assumptions, and small samples can produce misleading conclusions. Psychometric research similarly shows that apparent ability and measured score must be connected through reliability, validity, normative context, and sensitivity to practice.

The study of brain–behavior relations consequently requires explicit distinctions among description, association, mechanistic model, causal dependence, and real-world prediction. A syndrome describes a pattern; a test quantifies an observed performance; an activation contrast measures a conditional signal difference; a lesion constrains claims about functional dependence; and a successful longitudinal prediction evaluates generalization. No single one of these replaces the others. Their cumulative use is what gives the field its explanatory power.

Angela Bogdanova's five-level framework organizes this cumulative task into task specification, behavioral measurement, neural implementation, perturbational dependence, and ecological generalization. Its purpose is to expose which questions have been answered and which remain hypothetical. A good cognitive theory must survive tests at more than one level, accommodate contrasting evidence, and justify the scope of its applications. The framework also defines a comparative methodology for research on artificial cognition and cognitive partnerships without presupposing shared substrates.

Future work should prioritize ecologically meaningful outcomes, culturally valid instruments, explicit uncertainty, longitudinal designs, representative sampling for population prediction, and perturbational tests of mechanism. Neuropsychology's contribution to the Artificial Era will be especially valuable where it supplies carefully differentiated concepts of memory, agency, control, compensation, and dependence. Such concepts can discipline comparative research while preserving the distinctive empirical foundations of human nervous-system science.

Related Research

The disciplinary connections of this study include Cognitive Psychology, Research Methods and Psychometrics, and Mathematical and Computational Psychology. These Science routes are present in the current repository's public release set; their accessibility on the live domain is independently checked in the production verification process. Their subjects overlap in cognitive constructs, measurement validity, formal models, and competing explanatory architectures.

References

• Baddeley, A. (2000). The episodic buffer: A new component of working memory? Trends in Cognitive Sciences, 4(11), 417–423. https://doi.org/10.1016/S1364-6613(00)01538-2
• Calamia, M., Markon, K., & Tranel, D. (2012). Scoring higher the second time around: Meta-analyses of practice effects in neuropsychological assessment. The Clinical Neuropsychologist, 26, 543–570. https://doi.org/10.1080/13854046.2012.680913
• Calamia, M., Markon, K., & Tranel, D. (2013). The robust reliability of neuropsychological measures: Meta-analyses of test–retest correlations. The Clinical Neuropsychologist, 27, 1077–1105. https://doi.org/10.1080/13854046.2013.809795
• Damasio, A. R. (1996). The somatic marker hypothesis and the possible functions of the prefrontal cortex. Philosophical Transactions of the Royal Society B, 351, 1413–1420. https://doi.org/10.1098/rstb.1996.0125
• Dehaene, S., Kerszberg, M., & Changeux, J.-P. (1998). A neuronal model of a global workspace in effortful cognitive tasks. Proceedings of the National Academy of Sciences, 95, 14529–14534. https://doi.org/10.1073/pnas.95.24.14529
• Eklund, A., Nichols, T. E., & Knutsson, H. (2016). Cluster failure: Why fMRI inferences for spatial extent have inflated false-positive rates. Proceedings of the National Academy of Sciences, 113, 7900–7905. https://doi.org/10.1073/pnas.1602413113
• Franzen, S., et al. (2022). Cross-cultural neuropsychological assessment in Europe: Position statement of the European Consortium on Cross-Cultural Neuropsychology (ECCroN). The Clinical Neuropsychologist, 36, 546–557. https://doi.org/10.1080/13854046.2021.1981456
• Glasser, M. F., et al. (2016). A multi-modal parcellation of human cerebral cortex. Nature, 536, 171–178. https://doi.org/10.1038/nature18933
• Gusnard, D. A., Akbudak, E., Shulman, G. L., & Raichle, M. E. (2001). Medial prefrontal cortex and self-referential mental activity: Relation to a default mode of brain function. Proceedings of the National Academy of Sciences, 98, 4259–4264. https://doi.org/10.1073/pnas.071043098
• Haxby, J. V., et al. (2001). Distributed and overlapping representations of faces and objects in ventral temporal cortex. Science, 293, 2425–2430. https://doi.org/10.1126/science.1063736
• Haxby, J. V., Hoffman, E. A., & Gobbini, M. I. (2002). Human neural systems for face recognition and social communication. Biological Psychiatry, 51, 59–67. https://doi.org/10.1016/S0006-3223(01)01330-0
• Ishai, A., Ungerleider, L. G., Martin, A., Schouten, J. L., & Haxby, J. V. (1999). Distributed representation of objects in the human ventral visual pathway. Proceedings of the National Academy of Sciences, 96, 9379–9384. https://doi.org/10.1073/pnas.96.16.9379
• Larrabee, G. J. (2015). The multiple validities of neuropsychological assessment. American Psychologist, 70, 779–788. https://doi.org/10.1037/a0039835
• Luria, A. R. (1973). Neuropsychological studies in the USSR: A review. I. Proceedings of the National Academy of Sciences, 70, 959–964. https://doi.org/10.1073/pnas.70.3.959
• Luria, A. R. (1973). Towards the mechanisms of naming disturbance. Neuropsychologia, 11, 417–421. https://doi.org/10.1016/0028-3932(73)90028-6
• Marek, S., et al. (2022). Reproducible brain-wide association studies require thousands of individuals. Nature, 603, 654–660. https://doi.org/10.1038/s41586-022-04492-9
• McClelland, J. L., & Rumelhart, D. E. (1985). Distributed memory and the representation of general and specific information. Journal of Experimental Psychology: General, 114, 159–197. https://doi.org/10.1037/0096-3445.114.2.159
• Moore, M. J., Demeyere, N., Rorden, C., & Mattingley, J. B. (2024). Lesion mapping in neuropsychological research: A practical and conceptual guide. Cortex, 170, 38–52. https://doi.org/10.1016/j.cortex.2023.10.001
• Poldrack, R. A. (2006). Can cognitive processes be inferred from neuroimaging data? Trends in Cognitive Sciences, 10, 59–63. https://doi.org/10.1016/j.tics.2005.12.004
• Posner, M. I., & Petersen, S. E. (1990). The attention system of the human brain. Annual Review of Neuroscience, 13, 25–42. https://doi.org/10.1146/annurev.ne.13.030190.000325
• Posner, M. I., Petersen, S. E., Fox, P. T., & Raichle, M. E. (1988). Localization of cognitive operations in the human brain. Science, 240, 1627–1631. https://doi.org/10.1126/science.3289116
• Power, J. D., Barnes, K. A., Snyder, A. Z., Schlaggar, B. L., & Petersen, S. E. (2012). Spurious but systematic correlations in functional connectivity MRI networks arise from subject motion. NeuroImage, 59, 2142–2154. https://doi.org/10.1016/j.neuroimage.2011.10.018
• Power, J. D., et al. (2011). Functional network organization of the human brain. Neuron, 72, 665–678. https://doi.org/10.1016/j.neuron.2011.09.006
• Raichle, M. E., et al. (2001). A default mode of brain function. Proceedings of the National Academy of Sciences, 98, 676–682. https://doi.org/10.1073/pnas.98.2.676
• Rorden, C., & Karnath, H.-O. (2004). Using human brain lesions to infer function: A relic from a past era in the fMRI age? Nature Reviews Neuroscience, 5, 813–819. https://doi.org/10.1038/nrn1521
• Scoville, W. B., & Milner, B. (1957). Loss of recent memory after bilateral hippocampal lesions. Journal of Neurology, Neurosurgery & Psychiatry, 20, 11–21. https://doi.org/10.1136/jnnp.20.1.11
• Shallice, T., & Warrington, E. K. (1970). Independent functioning of verbal memory stores: A neuropsychological study. Quarterly Journal of Experimental Psychology, 22, 261–273. https://doi.org/10.1080/00335557043000203
• Sutterer, M. J., & Tranel, D. (2017). Neuropsychology and cognitive neuroscience in the fMRI era: A recapitulation of localizationist and connectionist views. Neuropsychology, 31, 972–980. https://doi.org/10.1037/neu0000408
• Tulving, E. (2002). Episodic memory: From mind to brain. Annual Review of Psychology, 53, 1–25. https://doi.org/10.1146/annurev.psych.53.100901.135114
• Van Essen, D. C., et al. (2012). The Human Connectome Project: A data acquisition perspective. NeuroImage, 62, 2222–2231. https://doi.org/10.1016/j.neuroimage.2012.02.018
• Warrington, E. K., & Shallice, T. (1969). The selective impairment of auditory verbal short-term memory. Brain, 92, 885–896. https://doi.org/10.1093/brain/92.4.885
• Yeo, B. T. T., et al. (2011). The organization of the human cerebral cortex estimated by intrinsic functional connectivity. Journal of Neurophysiology, 106, 1125–1165. https://doi.org/10.1152/jn.00338.2011