Randall Engle The Corsi Block-Tapping Test – Philip Corsi The Digit Span Task –

The quantification of human immediate memory capacity stands as one of the most enduring pursuits in cognitive science, experimental psychology, and clinical neuropsychology. From the earliest psychophysical investigations of sensory thresholds to modern computational architectures of executive control, the question of how many discrete units of information the human mind can actively maintain, manipulate, and retrieve over brief temporal intervals has shaped our foundational understanding of mental architecture. Within this broad empirical landscape, two paradigms have exercised singular influence over clinical practice and theoretical modeling: the Digit Span task, rooted in nineteenth-century psychoacoustics and standardized across iterations of the Wechsler intelligence scales, and the Corsi Block-Tapping Test, engineered within the surgical epilepsy wards of the Montreal Neurological Institute. These paradigms, though ostensibly straightforward clinical instruments, operationalize the primary sensory modalities through which human beings encode serial order: auditory-verbal phonology and dynamic visuospatial coordinates.

For several decades of the mid-twentieth century, the theoretical interpretation of these simple span tasks was dominated by structural, modular frameworks. The transition from the unitary memory models of early associationism to the dual-store architecture proposed by Richard Atkinson and Richard Shiffrin cast immediate memory as a passive, short-term buffer whose primary operational metric was discrete storage capacity. This conceptualization found natural alignment with the pioneering multi-component model introduced by Alan Baddeley and Graham Hitch in 1974, which explicitly fractionated this short-term store into modality-specific slave systems: the phonological loop, indexed effortlessly by the Digit Span task, and the visuospatial sketchpad, mirrored by the Corsi Block-Tapping Test. Under this prevailing view, span length reflected the physical boundary conditions of discrete, dedicated storage buffers, limited either by rapid temporal decay or item-based displacement within modular neural substrates.

However, this structuralist consensus was fundamentally challenged toward the close of the twentieth century by the theoretical and empirical work of Randall Engle and his collaborators. Through meticulous individual-differences research, structural equation modeling, and experimental cognitive paradigms, Engle demonstrated that simple span tasks—whether auditory or spatial—capture only a fraction of the cognitive variance necessary to predict complex, higher-order intellectual functioning. Engle reconceptualized Working Memory Capacity (WMC) not as the static capacity of domain-specific buffers, but as an active, domain-general system of executive attention: the capacity to maintain goal representations in an active state in the face of profound internal and external interference. This theoretical revolution forced a profound re-evaluation of both the Digit Span task and the Corsi Block-Tapping Test. This article provides an exhaustive examination of the historical evolution, psychometric structures, cognitive architectures, neuroanatomical substrates, and clinical diagnostic implications of these two classical paradigms, viewed through the transformative lens of Randall Engle’s executive attention theory.

1. Historical and Theoretical Foundations of Cognitive Span Paradigms

1.1 The Evolution from Short-Term Memory Buffers to Working Memory Systems

The shift from viewing immediate memory as a passive, transient buffer to characterizing it as an active, multi-layered working memory system represents one of the most significant paradigm shifts in modern cognitive psychology. Early conceptualizations, exemplified by the influential multi-store model formulated by Richard Atkinson and Richard Shiffrin (1968), posited a linear informational flow: environmental input entered brief sensory registers, transferred into a limited-capacity short-term store (STS), and, through rehearsal, was consolidated into the long-term store (LTS). Within this modal framework, the STS was conceived largely as a unitary, structural bottleneck. Its capacity limits—canonically characterized by George Miller in 1956 as the “magical number seven, plus or minus two”—were viewed as absolute physical properties of human mental processing, dictated by the number of discrete informational “slots” available for immediate maintenance.

Despite its intuitive elegance, the modal model proved structurally inadequate when confronted with clinical neuropsychological evidence. Patients with focal perisylvian lesions often displayed catastrophic deficits in auditory-verbal short-term memory—reducing their digit span to merely one or two items—yet they continued to demonstrate intact long-term memory consolidation, normal language comprehension, and preserved general reasoning. Such empirical anomalies led Alan Baddeley and Graham Hitch (1974) to dismantle the unitary short-term buffer, proposing in its place a dynamic, tripartite multicomponent working memory model. This architecture replaced the static storage container with an active system consisting of an attentional supervisory entity—the central executive—assisted by two specialized, domain-specific temporary maintenance systems: the phonological loop and the visuo-spatial sketchpad.

The introduction of the multicomponent model profoundly altered the interpretation of classical span tasks. Simple serial recall was no longer seen as tapping a homogenous cognitive container; instead, it was reinterpreted as the operational metric of specialized subsidiary stores designed to temporarily manipulate and preserve modality-congruent representations. Under this framework, capacity limitations ceased to be viewed merely as fixed item slots. Instead, they emerged as time-based and resource-dependent properties governed by the balance between rapid memory trace decay and the operational efficiency of active rehearsal mechanisms, such as sub-vocal articulatory cycling or dynamic spatial rehearsal.

1.2 The Emergence of Modality-Specific Testing in Neuropsychology

As cognitive psychology fractured immediate retention into specialized sub-components, clinical neuropsychology experienced a parallel imperative to develop testing paradigms sensitive to modality-specific neurofunctional architecture. Historically, neuropsychological assessment had been heavily weighted toward auditory-verbal instruments, primarily because verbal stimuli—such as strings of numbers, nonsense syllables, and word lists—were straightforward to standardize, present, and score in clinical settings. However, reliance on auditory-verbal instruments created severe diagnostic blind spots when evaluating individuals with primary language pathologies, expressive aphasias, left-hemisphere cerebrovascular accidents, or developmental language delays.

The clinical need to assess memory independently of linguistic competence became acute during the mid-twentieth century, particularly within burgeoning neurosurgical programs evaluating patients for unilateral temporal lobectomy. Clinicians and researchers recognized that verbal memory tests provided an incomplete index of cerebral integrity. Unilateral damage to right-hemisphere structures frequently left verbal memory performance entirely intact, masking profound neurocognitive deficits in non-verbal, spatial, and topographical orientation. A diagnostic dissociation was required: clinicians needed instruments that could quantify non-verbal, visuospatial immediate memory with the same psychometric precision, temporal pacing, and incremental difficulty that characterized classical verbal span metrics.

This methodological challenge demanded not merely non-verbal stimuli (such as static geometric shapes), but the preservation of temporal sequencing across space. True structural equivalence between verbal and non-verbal immediate memory paradigms required that the spatial test demand serial-order reconstruction. In everyday cognition, navigating an environment, remembering a trajectory, or imitating a sequence of physical actions requires encoding both “what” happened and the exact “temporal order” in which it occurred. Thus, developing a standardized apparatus capable of isolating dynamic spatiotemporal sequencing became a primary clinical imperative, leading directly to the innovation of physical, spatial block-tapping frameworks.

1.3 The Nexus of Philip Corsi, Randall Engle, and Classical Psychometrics

The intersection of Philip Corsi’s neuropsychological engineering, classical psychometrics, and Randall Engle’s modern cognitive frameworks represents a compelling intellectual evolution spanning five decades. In the early 1970s, under the supervision of Brenda Milner at the Montreal Neurological Institute (MNI), Philip Corsi devised his block-tapping paradigm to address the explicit need for a spatial analogue to the classic verbal Digit Span task. Corsi’s work successfully demonstrated that spatial serial recall was dissociable from verbal serial recall, mapping clean behavioral double dissociations onto left- versus right-hemisphere surgical lesions. For decades following, the Digit Span and Corsi Block-Tapping tests stood as the twin gold standards of modality-specific short-term storage assessment.

However, this classical psychometric paradigm—which treated span scores as pure reflections of regional brain function or domain-specific storage capacity—was radically disrupted by the work of Randall Engle and his laboratory beginning in the late 1980s and continuing through the 2000s. Engle noticed a persistent psychometric paradox: while simple span tasks like Digit Span and Corsi Blocks were sensitive to localized brain trauma and could separate verbal from spatial processing, they exhibited weak correlations with real-world cognitive performance, such as reading comprehension, fluid intelligence, logical reasoning, and academic achievement.

Engle revolutionized psychometrics by differentiating between simple span and complex span methodologies. Through extensive structural equation modeling and experimental manipulations, Engle demonstrated that simple spans measure primarily domain-specific coding, storage, and rehearsal strategies, whereas complex spans capture a domain-general capacity: executive attention. This theoretical conceptualization recontextualized both Corsi’s and Wechsler’s paradigms. In Engle’s architecture, simple spans are not obsolete; rather, their shared variance with complex tasks isolates the central executive machinery responsible for active goal maintenance amidst interference. Thus, tracing the historical arc from Corsi’s manual blocks to Engle’s automated, high-interference cognitive batteries illuminates how psychology moved from measuring the physical dimensions of the mind’s storage bins to quantifying the dynamic attentional mechanisms that govern human thought.

2. Philip Corsi and the Genesis of the Corsi Block-Tapping Test

2.1 Origins and Methodological Architecture in Montreal Neurological Institute

The Corsi Block-Tapping Test was introduced by Philip Michael Corsi in his 1972 doctoral dissertation at McGill University, titled “Human Memory and the Medial Temporal Region of the Brain.” Working within the legendary neurosurgical milieu of Brenda Milner, Corsi’s primary research objective was to delineate the functional consequences of unilateral temporal-lobe resections in epileptic patients, particularly with respect to the lateralization of memory functions. While Milner and her colleagues had demonstrated that left temporal lobectomy reliably impaired the learning and retention of verbal material, the quantified behavioral consequences of right temporal lobectomy had proven elusive when measured by conventional psychometric tests.

To overcome this limitation, Corsi designed a physical testing board engineered to prevent verbalization and verbal coding strategies. The original apparatus consisted of nine identical wooden cubes (measuring 1.5 inches or 38 mm on each side) affixed in a fixed, non-symmetrical, pseudo-random array across a rectangular wooden baseboard (typically 9 by 11 inches). The critical design element of Corsi’s layout was its deliberate lack of geometric regularity: the blocks were distributed such that they did not form straight lines, squares, circles, or easily identifiable geometric shapes, thereby frustrating any intuitive alphanumeric numbering or verbal mnemonic categorization by the participant.

The physical blocks were plain on the side facing the patient, presenting an entirely homogenous visual field. On the side facing the examiner, however, the blocks were subtly labeled with numbers 1 through 9. This arrangement allowed the examiner to administer complex, pre-standardized spatial sequences smoothly without hesitating or signaling the target blocks through visual gaze or physical postural cues. The manual administration protocol was strictly timed: the examiner tapped a predetermined sequence of blocks at a constant cadence—precisely one block per second—subsequently signaling the patient to reproduce the identical tapping sequence immediately upon completion.

The testing protocol followed a graded span procedure directly adapted from verbal memory testing. Sequences began with simple paths of two or three blocks and incrementally increased in length. If a participant successfully reproduced at least one sequence of a given length across two or three trials, the sequence length was incremented by one block. Testing discontinued once the participant failed a predetermined proportion of trials at a given sequence length (typically failing two out of three trials). Corsi’s baseline scoring derived a “spatial span”—the maximum sequence length reproduced correctly—establishing a reliable non-verbal counterpart to the acoustic-verbal spans of mainstream psychometrics.

2.2 Targeted Cognitive Constructs: The Visuospatial Sketchpad

From the perspective of cognitive architecture, the Corsi Block-Tapping Test targets the operational mechanics of the visuospatial sketchpad (VSSP). While early formulations of Baddeley’s working memory model treated the VSSP as a relatively unitary store for non-phonological representations, subsequent theoretical refinement—most notably driven by the work of Robert Logie (1995)—demonstrated that the sketchpad must be fractionated into at least two functionally and anatomically dissociable subcomponents: the visual cache and the inner scribe.

The visual cache serves as a passive perceptual store responsible for retaining static visual details, such as color, shape, pattern complexity, and visual texture. In contrast, the inner scribe is an active, dynamic rehearsal mechanism responsible for retaining spatial sequences, movement trajectories, and the temporal transitions between spatial coordinates. The Corsi Block-Tapping Test places minimal demand on the visual cache; because all nine blocks are physically identical in color, size, and material, object-identity information offers zero discriminative value for task success. Instead, the task relies almost exclusively on the inner scribe.

Executing a Corsi trial requires the participant to perform three distinct mental operations:

• Spatial Coordinate Encoding: Rapidly mapping the absolute spatial location of each tapped block relative to the physical boundaries of the testing board and bodily egocentric coordinates.
• Dynamic Vector Tracking: Encoding the directional motion vectors connecting one block to the next, translating discrete visual flashes or tapping movements into a continuous spatiotemporal trajectory.
• Motor Sequence Rehearsal: Actively maintaining the spatiotemporal path during the retention interval via covert motor programming, utilizing neural systems shared with oculomotor saccade planning and manual reaching.

Because the inner scribe is fundamentally linked to motor planning, spatial sequence retention is profoundly vulnerable to spatial and kinesthetic interference. When participants are required to execute continuous spatial tapping (such as tapping four corners of a table in a circular pattern) during the retention interval of a Corsi task, performance drops precipitously. Conversely, concurrent articulatory suppression maneuvers (such as repeating “the, the, the” out loud) have virtually no impact on basic Corsi span in neurologically intact adults who do not rely on covert verbal numbering strategies, reinforcing the construct validity of the Corsi test as an index of dynamic spatial working memory.

2.3 Clinical Utility and Sensitivity to Neuroanatomical Lesions

The enduring clinical value of the Corsi Block-Tapping Test lies in its remarkable diagnostic sensitivity to unilateral, lateralized cerebral trauma, particularly within the right hemisphere. Corsi’s original 1972 findings provided definitive evidence that patients who had undergone right temporal lobectomy—especially those with extensive resections of the right hippocampal formation and parahippocampal gyrus—exhibited severe impairments in reproducing spatial sequences that exceeded their immediate spatial span, whereas their performance on equivalent verbal tasks remained indistinguishable from healthy controls.

Subsequent neurosurgical and neuropsychological investigations broadened the diagnostic profile of the Corsi test. While the hippocampus is critical for binding complex spatial relationships and retaining them over supraspan delays, immediate spatial sequence reproduction (the span itself, typically 5 items in healthy adults) relies heavily on a distributed network encompassing the right posterior parietal cortex, the right dorsolateral prefrontal cortex, and interconnected frontostriatal circuits. Patients suffering from acute ischemic strokes in the distribution of the right middle cerebral artery reliably demonstrate marked reductions in Corsi span scores.

Furthermore, the Corsi Block-Tapping Test serves as an indispensable instrument in diagnosing and characterizing unilateral spatial neglect and constructional apraxia. Patients with right parietal lesions exhibiting hemispatial neglect frequently fail Corsi sequences not because of an intrinsic mnemonic failure, but because their attentional orienting systems fail to register taps occurring within the contralesional (left) hemispace of the testing board. By systematically comparing sequence reproduction across trials where taps are isolated to the right half of the board versus trials crossing the midline into the left hemifield, clinicians can effectively decouple pure spatial memory decay from primary hemispatial attentional deficits.

3. The Digit Span Task: Psychometric Structure and Phonological Mechanics

3.1 Historical Lineage from Jacobs and Ebbinghaus to the Wechsler Scales

The Digit Span task possesses the longest continuous clinical and experimental lineage of any instrument in cognitive psychology. The paradigm was conceived in 1887 by the London schoolmaster Joseph Jacobs, who sought a quantitative metric to measure the “prehension span”—the pure capacity of mental apprehension—in school-age children. Jacobs recognized that individual letters and words suffered from significant linguistic confounds: differential vocabulary exposure, semantic associations, and varying familiarity across socio-economic strata. Digits, by contrast, represented an ideal psychometric compromise: they were universally recognized, acoustically discrete, monosyllabic (in the English language, with the exception of ‘seven’), and presented equal familiarity to all test subjects.

Shortly thereafter, Hermann Ebbinghaus adapted span principles to explore the mathematical properties of memory retention and savings during his foundational work on the forgetting curve. The diagnostic value of the digit span metric was formalized when Alfred Binet and Théodore Simon incorporated forward digit repetition into their 1905 intelligence scale. Binet recognized that the capacity to immediately repeat a sequence of numbers scaled monotonically across normative childhood development, providing an anchor for establishing mental age.

The institutionalization of the Digit Span task into adult clinical psychometrics occurred through David Wechsler. Wechsler incorporated the Digit Span subtest into the Wechsler-Bellevue Intelligence Scale in 1939, an architecture that persisted through subsequent iterations: the WAIS, WAIS-R, WAIS-III, WAIS-IV, and the latest WAIS-V batteries. Over these decades, psychometric administration standards underwent rigorous formalization. Standardized testing dictates that digits must be delivered at an unhurried, perfectly uniform rate of one digit per second, with the examiner’s pitch dropping gently on the final digit of the sequence to avoid rhythmic grouping artifacts. The standardization of delivery speed, neutral vocal inflection, and explicit discontinuation criteria established the Digit Span task as the global reference standard for immediate auditory-verbal recall.

3.2 Phonological Loop Dynamics and Articulatory Suppression

From the vantage point of cognitive mechanisms, the Forward Digit Span operates primarily through the phonological loop within Baddeley’s multicomponent framework. The phonological loop comprises two tightly integrated sub-mechanisms: a passive phonological store and an active articulatory rehearsal process.

When digits are spoken by an examiner, acoustic signals gain immediate, obligatory access to the phonological store. Within this store, digits are represented as transient acoustic traces that are exceptionally vulnerable to temporal decay, typically fading completely within 1.5 to 2 seconds unless refreshed. This refresh operation is executed by the articulatory rehearsal process, commonly understood as “inner speech.” By silently repeating the numerical sequence in cyclical succession, the articulatory mechanism retrieves traces from the phonological store and re-injects them into the buffer, arresting passive decay.

The mechanistic validity of this architecture is confirmed through several robust empirical phenomena:

• The Word-Length Effect: Memory span is an inverse function of the spoken duration of the words presented. Individuals can consistently recall more one-syllable words than multi-syllable words because shorter words can be recited faster within the 2-second decay threshold. In languages where digits are polysyllabic (e.g., Welsh compared to English), baseline digit spans are systematically lower, directly reflecting articulatory rehearsal speed.
• The Acoustic Similarity Effect: Sequences of phonologically similar items (e.g., B, C, D, P, T) are significantly harder to recall in correct serial order than phonologically distinct sequences. The shared phonetic characteristics generate catastrophic trace interference within the phonological store.
• Articulatory Suppression: If a participant is instructed to continuously vocalize an irrelevant syllable (such as repeating “cola, cola, cola”) during presentation and retention, the articulatory loop is mechanically blocked. Under articulatory suppression, the forward digit span drops dramatically, reducing performance to the raw, un-rehearsed lifespan of the passive phonological trace.

3.3 Structural Disparities Between Forward and Backward Digit Span

Although historical psychometric batteries routinely pooled Forward Digit Span and Backward Digit Span into a single aggregated score, cognitive psychology and clinical psychometrics treat them as structurally, computationally, and neurologically disparate tasks. Forward Digit Span is predominantly a measure of passive phonological reception and temporal storage capacity. The participant functions essentially as an echoic transducer: the sequence is heard, temporarily held in the phonological store, kept alive via subvocal cycling, and reproduced in a linear, un-manipulated read-out.

Backward Digit Span, conversely, imposes a heavy computational tax on the central executive. To repeat a series of digits in reverse order (e.g., hearing “5 – 8 – 2 – 9” and outputting “9 – 2 – 8 – 5”), the participant cannot simply read out the temporal trace stored in the phonological loop. Doing so would produce the forward sequence and disrupt the reverse goal state. Instead, the individual must maintain the original phonological sequence in an active state while systematically decoupling the items from their initial serial positions.

This operation introduces substantial cognitive resistance: the participant must mentally suppress the dominant temporal associations formed during the presentation, sequentially isolate the terminal digit, output it via the motor articulatory apparatus, and simultaneously prevent the remaining, non-retrieved items from decaying or succumbing to proactive interference. Factor-analytic studies consistently reveal that while Forward Digit Span loads almost exclusively on simple short-term storage factors, Backward Digit Span loads robustly onto Working Memory and Central Executive factors alongside complex tasks like mental arithmetic, letter-number sequencing, and matrix reasoning.

4. Randall Engle and the Executive Attention Theory of Working Memory Capacity

4.1 The Paradigm Shift: From Memory Storage to Controlled Attention

In the late 1980s and 1990s, the field of cognitive psychology confronted an empirical crisis regarding the real-world predictive validity of short-term memory tests. For decades, simple spans had been presumed to capture the core capacity limitations of human intelligence. Yet, study after study demonstrated that simple Forward Digit Span and Corsi Block spans exhibited strikingly low or inconsistent correlations with real-world intellectual benchmarks, including general fluid intelligence (Engle, Kane, & Tuholski, 1999), scholastic aptitude scores, language comprehension capabilities, and complex problem-solving abilities.

Randall Engle and his research group resolved this paradox by fundamentally redefining the construct of Working Memory Capacity (WMC). Engle dismantled the long-standing assumption that working memory capacity reflects the maximum number of items that can be maintained in an active structural buffer or temporary container. Instead, Engle formulated the Executive Attention Theory of Working Memory. Under this theoretical framework, WMC is not a memory construct in the traditional sense; it is a manifestation of controlled, executive attention.

Engle formalized WMC as the cognitive capability to actively maintain goal-relevant representations—such as operational rules, target items, or behavioral constraints—in the presence of internal or external distraction, and to actively suppress potent, irrelevant habits, prepotent responses, and interference. In Engle’s architecture, working memory is conceptualized as:

1. The vast, permanent structural architecture of long-term memory representations that have been raised above an activation threshold by associative cues.
2. A strictly limited, focus-of-attention mechanism governed by the prefrontal cortex, which exerts continuous top-down control to bias activation toward task goals and actively inhibit irrelevant distractors.

Thus, when an individual succeeds on a demanding cognitive challenge, their success is not governed by possessing a physically larger storage bin. Rather, it reflects their ability to maintain attentional focus over time, rapidly recover from attentional capture, and shield fragile mental representations from proactive interference.

4.2 Complex Span Paradigms vs. Simple Span Paradigms

To capture this newly identified executive attention construct empirically, Engle, Daneman, Carpenter, and their colleagues developed the complex span methodology. Unlike simple span tasks, which present an uninterrupted sequence of items to be immediately retrieved, complex span tasks explicitly interleave a secondary, processing-intensive task between the presentations of each memory item.

Prominent variations of Engle’s complex span paradigms include:

• Operation Span (O-Span): Participants must solve a mathematical equation (e.g., Is (4 × 2) + 1 = 9?) and immediately confirm its accuracy, after which an arbitrary letter or word is presented for subsequent recall. This cycle repeats across sequence lengths ranging from two to seven items, culminating in serial recall of the stored letters.
• Reading Span (R-Span): Participants read aloud a grammatically complex sentence, judge its semantic coherence, and encode a terminal word for subsequent serial reproduction.
• Symmetry Span (S-Span): The spatial analogue of O-Span, requiring participants to evaluate the vertical symmetry of a pixelated black-and-white grid before viewing a colored square illuminated in a dynamic spatial matrix.

The diagnostic power of complex span tasks lies in the functional role of the interleaved processing task. By forcing the participant to divert their conscious attention to verifying an equation or reading a sentence, the processing component prevents the use of automatic, continuous subvocal rehearsal (in the verbal domain) or continuous motor tracking (in the spatial domain). Once simple rehearsal strategies are blocked, the participant must rely on executive attention to periodically switch focus back to the decaying memory traces, retrieve them from activated long-term memory, and shield them from the severe proactive interference generated by previous processing trials.

Through large-scale latent variable analyses and structural equation modeling involving thousands of subjects, Engle’s laboratory proved that the latent variance extracted from complex span tasks accounts for anywhere between 30% and 50% of the variance in general fluid intelligence ($G_f$). Simple span tasks, by contrast, load almost entirely on task-specific, domain-restricted storage factors that fail to predict higher-order cognition once the shared variance of executive attention is statistically removed.

4.3 Proactive Interference, Goal Maintenance, and Cocktail Party Phenomena

Engle’s empirical investigations into the mechanisms of executive attention illuminated three foundational pillars: resistance to proactive interference, active goal maintenance, and selective attentional filtering. Among these, the suppression of proactive interference (PI) stands as the primary cognitive operational mechanism that separates high-WMC individuals from their low-WMC peers.

Proactive interference occurs when previously encoded information disrupts the encoding and retrieval of newly acquired information. In a classic simple span test, trials occur rapidly in succession. For low-WMC individuals, traces from Trial 1, Trial 2, and Trial 3 remain partially activated in memory, progressively degrading performance on subsequent trials. Kane and Engle (2000) demonstrated that when proactive interference is explicitly minimized (e.g., on the very first trial of an experiment, or when semantic categories are shifted), individuals with low WMC perform almost as well as individuals with high WMC. It is only when PI accumulates across multiple trials that the performance gap widens drastically. High-WMC individuals utilize controlled attention to actively suppress outdated temporal traces, isolating the current target list from previously learned material.

This capability for robust goal maintenance in high-interference environments generalizes far beyond traditional memory tasks. In a classic demonstration of attentional control, Conway, Cowan, and Engle (2001) tested participants in the classic “cocktail party” dichotic listening paradigm. Participants listened to two divergent audio streams piped into separate ears and were instructed to attend solely to the message in one ear while shadowing it (repeating it aloud). When the subject’s own name was unexpectedly whispered into the unattended ear, 65% of individuals with low WMC noticed their name, failing to shield their primary task goal from salient distraction. In striking contrast, only 20% of individuals with high WMC detected their name.

High WMC does not reflect a heightened perceptual capacity to register external sensory stimuli; rather, it reflects a superior ability to block out irrelevant information and preserve fidelity over the primary behavioral objective. This same dynamic explains why high-WMC individuals perform significantly better on the Stroop task (overcoming prepotent reading reflexes) and anti-saccade tasks (suppressing the biological reflex to look toward a flickering peripheral visual stimulus, instead looking in the opposite direction). Working memory capacity, as redefined by Engle, is the metric of mental resistance against distraction.

5. Structural and Psychometric Comparison: Corsi Block-Tapping vs. Digit Span

5.1 Modality Dissociations: Spatial Coordinates vs. Phonological Codes

The fundamental theoretical division separating the Corsi Block-Tapping Test and the Digit Span task lies in their sensory modalities: the former is grounded in dynamic egocentric spatial coordinates, whereas the latter operates entirely upon sequential phonological acoustic tokens. This difference creates distinct cognitive processing requirements across the two paradigms.

In the Digit Span task, the input consists of categorical, highly practiced linguistic symbols. Human adults have engaged in decades of acoustic, phonological, and semantic associations with digits. Each digit represents a bounded categorical token; there is no ambiguity regarding whether an examiner uttered a “three” or a “four.” As a result, the computational burden on the participant during Digit Span is not identification of the stimulus, but preserving the exact temporal order of the sequence. Acoustic tokens are rapidly converted into phonemes, mapped onto motor programs within the speech cortex, and maintained through cyclic vocal or subvocal rehearsal loops.

Conversely, the Corsi Block-Tapping Test requires the rapid perceptual parsing of continuous, analog spatial metrics. The blocks do not carry inherently distinct categorical identities; they are visually identical wooden or plastic cubes. The participant must encode each event as a coordinate vector in physical space relative to other blocks, the frame of the board, and their own body axis. Serial order in the Corsi task cannot be easily offloaded to automatic speech motor loops; instead, it must be represented through dynamic spatiotemporal trajectories, sequence-dependent motor planning, and visual tracking networks.

This modality distinction produces classic double dissociations in experimental paradigms:

Cognitive Dimension	Digit Span Paradigm	Corsi Block-Tapping Paradigm
Primary Coding Modality	Acoustic, phonological, symbolic	Spatial coordinates, continuous motor vectors
Rehearsal Mechanism	Subvocal articulatory loop	Oculomotor planning, dynamic inner scribe
Selective Interference Sensitivity	Acoustic similarity, articulatory suppression	Visuospatial tracking, dynamic motor tapping
Primary Cerebral Lateralization	Left hemisphere perisylvian networks	Right hemisphere frontoparietal networks
Mean Healthy Adult Span Length	$7 \pm 2$ items (digits)	$5 \pm 1$ items (spatial locations)

The difference in mean span length between the two tasks—healthy adults average approximately 7 digits on Forward Digit Span, but only 5 blocks on Forward Corsi—highlights the distinct biological constraints of the underlying substrates. Phonological rehearsal can cycle multiple monosyllabic items within a ~2-second window. In contrast, dynamic spatial encoding requires traversing spatial distances, coordinating saccades, and activating motor planning sequences, operations that take longer per item and compress the absolute number of items that can be maintained before trace degradation occurs.

5.2 Psychometric Reliability, Validity, and Floor/Ceiling Effects

From a psychometric perspective, both the Digit Span and Corsi Block-Tapping tests demonstrate solid test-retest reliability, though each exhibits specific measurement constraints depending on administration modality and sample demographics. Standardized implementations of the WAIS Digit Span consistently yield split-half and test-retest reliability coefficients exceeding $r = 0.85$, reflecting the standardized nature of acoustic timing, discrete item scoring, and strict discontinuation criteria.

The manual Corsi Block-Tapping Test, however, exhibits slightly more variable reliability profiles across published clinical literature, with test-retest coefficients typically ranging between $r = 0.70$ and $r = 0.82$. This slight depression in psychometric stability is largely attributable to examiner variance during manual administration. Small discrepancies in the speed of the examiner’s hand movement, subtle variations in inter-tap pauses, accidental physical hesitation over candidate blocks, and unconscious directional eye-gaze shifts introduce uncontrolled measurement noise that can artificially inflate or depress a patient’s spatial span score. The advent of computerized, touch-screen implementations (such as e-Corsi and PEBL implementations) has largely corrected this vulnerability, elevating reliability coefficients to parity with standardized verbal tests.

Regarding construct validity, both paradigms demonstrate clear convergent validity with other metrics within their respective cognitive domains. Digit Span correlates strongly with verbal list learning, auditory consonant trigrams, and reading speed. The Corsi test correlates robustly with visual pattern reproduction, mental rotation paradigms, and maze navigation tasks. However, both paradigms suffer from noticeable ceiling effects when administered to high-functioning, healthy young adult populations. Because simple span tasks feature relatively low cognitive resistance, high-ability individuals quickly reach the upper bound of the test, relying on idiosyncratic grouping or visualization strategies to maximize performance. Furthermore, their discriminant validity against general fluid intelligence is notoriously weak; when evaluated via structural equation modeling, the standalone correlation between simple Corsi or Digit spans and matrix reasoning tasks rarely exceeds $r = 0.25$, demonstrating that simple storage tests alone cannot adequately capture higher-order intellectual ability.

5.3 Heuristic Strategies: Verbalization of Corsi and Visualization of Digits

A persistent methodological challenge in administering both the Corsi Block-Tapping Test and the Digit Span task is the human brain’s natural tendency toward cross-modal recoding. The human cognitive apparatus is opportunistic; when presented with an arbitrary task, participants routinely deploy cognitive strategies that translate difficult representations from one sensory modality into another where they possess greater facility or strategic expertise.

In the Corsi Block-Tapping Test, participants frequently attempt to convert the spatial array into a verbal schema. A common heuristic involves mentally numbering the blocks 1 through 9 (e.g., assigning 1 to the top-left block, 9 to the bottom-right block) or inventing descriptive spatial narratives (e.g., “top corner, down to middle, bounce to the right edge”). Once translated into an alphanumeric code, the participant no longer relies on the inner scribe or spatial tracking networks; instead, they offload the task onto the phonological loop, reciting the invented numbers subvocally. To confirm whether a participant is relying on spatial mechanisms or covert verbal recoding, experimental neuropsychologists introduce articulatory suppression. If reciting meaningless syllables out loud severely degrades a subject’s Corsi performance, the examiner can deduce that the individual was relying on cross-modal verbal strategies rather than true spatial representation.

Conversely, during the Digit Span task, individuals frequently employ visual and spatial heuristics to bypass the limits of phonological rehearsal. A prevalent strategy among mnemonic practitioners and high-span individuals involves projecting spoken digits onto a mental number line or visualizing them inscribed across a physical matrix (such as an imagined keypad). Others engage in structural chunking, transforming linear strings of numbers into integrated semantic units (e.g., recoding the sequence “1, 9, 4, 5” from four discrete phonological items into a single temporal-semantic chunk: the historical year “1945”). This strategic chunking compresses the total information load, artificially elevating span performance without reflecting an expansion of raw working memory capacity.

6. Engle’s Critique of Simple Span Tasks: Re-evaluating Corsi and Digit Span

6.1 Simple Storage vs. Executive Control: The Empirical Evidence

The critique mounted by Randall Engle and his associates against the clinical reliance on simple span tasks represented an empirical watershed in cognitive psychology. Engle did not dispute that the Digit Span and Corsi Block-Tapping tasks measured something real; his critique focused instead on what they measured. For decades, researchers had operated under the implicit assumption that Forward Digit Span and Forward Corsi were micro-scale proxies for the overall computational capacity of the human mind. Engle exposed the psychometric limitations of this assumption.

In a series of landmark studies, Engle, Tuholski, Laughlin, and Conway (1999) administered a comprehensive battery of simple storage spans (Digit Span, Word Span, Letter Span) and complex spans (Operation Span, Reading Span, Counting Span) to large normative cohorts, analyzing the resulting covariance structures through confirmatory factor analysis and structural equation modeling (SEM). The empirical findings were definitive:

Structural Path in SEM Framework	Observed Correlation / Path Weight ($\beta$)	Theoretical Implication
Simple Spans $\rightarrow$ Short-Term Memory (STM) Factor	$\beta \approx 0.80 – 0.95$	Simple spans load predominantly on domain-specific storage and rehearsal.
Complex Spans $\rightarrow$ Working Memory Capacity (WMC) Factor	$\beta \approx 0.85 – 0.95$	Complex spans successfully isolate domain-general executive attention.
STM Factor $\rightarrow$ Fluid Intelligence ($G_f$)	$\beta \approx 0.15 – 0.25$ (Weak / Non-significant)	Pure storage capacity does not drive high-level reasoning or problem-solving.
WMC Factor $\rightarrow$ Fluid Intelligence ($G_f$)	$\beta \approx 0.60 – 0.75$ (Strong / Predictive)	Executive attention and interference control account for the shared variance with fluid intelligence.

These findings provided clear empirical evidence decoupling simple storage from working memory capacity. While the Digit Span task was effective at detecting transient phonological storage breakdown, its variance—once stripped of executive attention—had virtually no relationship with whether a person could comprehend a complex legal text, execute advanced mathematical reasoning, or solve novel matrix puzzles. The Corsi Block-Tapping Test yielded identical outcomes in the visuospatial domain: simple spatial sequence reproduction loaded heavily on low-level spatial perception and motor execution factors, but showed minimal predictive utility for higher-order spatial reasoning or abstract fluid cognition.

6.2 Modifying Simple Spans: The Introduction of Cognitive Resistance

To examine why complex spans predict fluid intelligence whereas simple spans often fail, Engle’s laboratory conducted experiments designed to inject “cognitive resistance” directly into traditional span formats. The primary factor distinguishing simple tasks from complex tasks is that the latter prevent the participant from using automatic, domain-specific rehearsal to maintain the targets.

When an individual takes a traditional Forward Digit Span test, the task encourages a continuous loop of inner speech: “4-7-1-9, 4-7-1-9.” This strategy requires minimal prefrontal executive control; it is sustained largely through automatic motor-perceptual feedback loops between Broca’s area and the left superior temporal gyrus. The task demands little conscious mental focus, requires zero manipulation of the targets, and occurs under low-interference conditions. The same dynamic applies to the Forward Corsi task: participants maintain the spatial sequence through continuous, covert eye movements or motor trajectory preparation. The mind is not required to fight off competing goals or resolve severe informational conflict.

To demonstrate that attentional resistance is the key operational variable, Engle and his colleagues introduced modified span paradigms. In these designs, participants were given simple lists of digits or spatial locations, but were subjected to an interleaved processing demand—such as verifying symmetry, solving a simple arithmetic problem, or resisting a potent perceptual distractor—between each target presentation. The results were clear: as soon as cognitive resistance was introduced, preventing automatic rehearsal and inducing proactive interference, performance on the simple digit and spatial span tasks dropped to levels comparable to complex span tasks, and their correlations with fluid intelligence and executive functioning rose dramatically.

Backward transformations (such as Backward Digit Span and Backward Corsi) serve as a partial, real-world clinical example of this transformation. Because the participant must actively reverse the sequence rather than merely echo it, the backward modality introduces cognitive resistance. The backward task disrupts automatic linear rehearsal, forcing the frontal executive network to intervene to maintain the goal state, manipulate the informational sequence, and suppress proactive interference from the forward trace.

6.3 The Unitary vs. Non-Unitary Debate on Executive Attention

The theoretical divergence between Baddeley’s multi-component model and Engle’s executive attention theory ignited an important debate in cognitive psychology: Is executive working memory capacity fundamentally unitary (domain-general) or non-unitary (domain-specific)?

Baddeley’s model has traditionally leaned toward a modular, domain-specific architecture. It posits distinct, domain-specific storage buffers (the phonological loop and visuospatial sketchpad) overseen by a central executive. However, Baddeley’s early formulations rarely operationalized the specific mechanisms of this central executive in predictive psychometric models. In contrast, Randall Engle championed a strictly domain-general conceptualization of executive control. Engle argued that while the representations held in memory are undoubtedly domain-specific (taking the form of phonological codes in the left hemisphere or spatial coordinates in the right hemisphere), the attentional control mechanism that preserves those representations amidst distraction is entirely domain-general.

To evaluate this hypothesis, Engle, Kane, and their colleagues constructed cross-domain latent variable models. They tested whether verbal working memory capacity (derived from Operation Span and Reading Span) and spatial working memory capacity (derived from Symmetry Span and Rotation Span) loaded on separate, domain-specific executive factors or on a single, overarching executive attention factor. The structural equation modeling provided strong support for the domain-general view:

• The correlation between the latent executive factor extracted from verbal complex spans and the latent executive factor extracted from spatial complex spans routinely approached unity ($r ge 0.85 – 0.95$).
• The executive attention factor independently predicted fluid intelligence across both verbal and spatial reasoning tests, completely mediating the relationship between short-term storage and higher-order cognition.

Under Engle’s framework, re-evaluating the Corsi Block-Tapping Test and the Digit Span task reveals that they are not isolated cognitive silos. Instead, they represent two sensory windows into the human cognitive architecture. When simple spans are administered in forward mode, their divergence reflects domain-specific modular representations. But as task difficulty increases, interference accumulates, and active sequence manipulation is required, both the Corsi and Digit Span tasks converge upon a shared, domain-general prefrontal executive network whose primary operational metric is the controlled deployment of visual and phonological attention.

7. Neuroanatomical and Neuroimaging Correlates

7.1 Neural Substrates of the Digit Span Task

Decades of structural lesion mapping and functional neuroimaging (PET, fMRI) investigations have delineated the neural circuits that support the execution of the Digit Span task. The neuroanatomical signature of Forward Digit Span is localized predominantly within the left cerebral hemisphere, following the classical left-lateralized perisylvian language architecture.

The passive storage component of the phonological loop relies heavily on the left posterior parietal cortex, specifically the left supramarginal gyrus (Brodmann Area 40). Functional imaging shows that when acoustic digit sequences are presented and passively retained, blood-oxygen-level-dependent (BOLD) signals rise significantly across this inferior parietal region. Neurological patients with focal infarctions restricted to the left supramarginal gyrus present with a classic syndrome: profoundly depressed forward digit spans (often limited to 1–3 items) despite fully preserved motor speech, normal lexical comprehension, and intact long-term memory consolidation.

Conversely, the active, articulatory rehearsal mechanism engages the anterior motor and premotor components of the left hemisphere. The primary driver of subvocal cycling is Broca’s area (left inferior frontal gyrus, Brodmann Areas 44 and 45), working in concert with the left premotor cortex (Brodmann Area 6), the supplementary motor area (SMA), and the bilateral anterior insula. When participants maintain a digit list via inner speech, this fronto-parietal articulatory network mirrors the neural activation observed during overt motor vocalization, but without downstream peripheral motor unit contraction.

When the task switches from Forward to Backward Digit Span, functional neuroimaging reveals a profound shift in neural recruitment. The unilateral left-hemisphere network transitions into an expansive, bilateral frontoparietal cognitive control system. The most pronounced activation occurs within the left and right dorsolateral prefrontal cortex (DLPFC, Brodmann Areas 9 and 46), accompanied by the bilateral intraparietal sulcus (IPS) and the dorsal anterior cingulate cortex (ACC). The recruitment of the bilateral DLPFC directly reflects the requirement for central executive control: updating working memory representations, sequentially reorganizing the phonological traces, and suppressing the forward echoic trace to permit reverse serial retrieval.

7.2 Neural Substrates of the Corsi Block-Tapping Test

The Corsi Block-Tapping Test recruits a neural network that is distinctly right-lateralized in healthy, neurologically intact adults, mapping onto systems dedicated to spatial attention, oculomotor planning, and motor trajectory tracking.

The primary cortical region supporting the encoding and short-term retention of spatial coordinates in the Corsi paradigm is the right posterior parietal cortex, centered around the superior parietal lobule (SPL, Brodmann Area 7) and the intraparietal sulcus (IPS). The intraparietal sulcus constructs dynamic, egocentric spatial maps that locate targets relative to the body. Functional MRI studies demonstrate that during the encoding of Corsi sequences, BOLD signals within the right IPS scale directly with sequence length: as the number of tapped blocks increases from two to six, parietal metabolic activation increases proportionally until the participant’s spatial span limit is exceeded.

The spatial rehearsal mechanism of the Corsi test relies on the frontal eye fields (FEF, Brodmann Area 8), the supplementary eye fields (SEF), and the right premotor cortex. Because the inner scribe maintains spatial sequences through covert shifts of spatial attention, this rehearsal process directly engages the neural circuitry of oculomotor saccade planning. Even when a participant remains physically motionless during the retention interval, covert gaze planning continuously cycles through the target coordinates, activating the frontal eye fields and the superior colliculus.

The retrieval and motor reproduction phase involves the basal ganglia (particularly the caudate nucleus and putamen) and the cerebellum, which translate spatial representations into precise motor reaching trajectories. Regarding medial temporal structures: while immediate spatial span (up to 4–5 blocks) can be maintained via frontoparietal networks alone, spatial sequences that exceed the span limit or require retention over supraspan delays depend on the integrity of the right hippocampus and adjacent parahippocampal cortex, which bind disparate spatial coordinates into cohesive relational maps in long-term memory.

7.3 Engle’s Prefrontal Cortex (PFC) Model of Executive Attention

The neuroanatomical foundations of Randall Engle’s executive attention theory shift the empirical focus from modality-specific posterior sensory areas to the supervisory control networks of the frontal lobes. In Engle’s framework, Working Memory Capacity is mediated by the integrity and functional connectivity of the prefrontal cortex (PFC), specifically its interactions with the dorsal anterior cingulate cortex (dACC) and the frontoparietal control network (FPN).

Within this model, the dorsolateral prefrontal cortex (DLPFC) does not function as an informational storage bin. It does not contain specialized neurons dedicated to storing phonemes or spatial coordinates. Instead, the DLPFC operates as an attentional bias engine. It sends robust, top-down excitatory signals to posterior cortical areas (such as the supramarginal gyrus for phonemes or the intraparietal sulcus for spatial points) to maintain their activation levels, while simultaneously sending inhibitory signals to suppress alternative representations that could induce proactive interference.

The dorsal anterior cingulate cortex (dACC) plays a critical role in conflict detection within Engle’s neurocognitive architecture. When a participant takes a complex span task, the dACC detects the informational conflict that arises between the current target items and the lingering proactive interference from previous trials. Once conflict is signaled, the dACC recruits the DLPFC to increase attentional control, stabilize the task goal, and suppress competing memory traces.

Functional neuroimaging studies by Engle’s collaborators (e.g., Kane & Engle, 2002; Burgess et al., 2011) demonstrate that high-WMC individuals display distinct neural efficiency profiles compared to low-WMC peers. Under high interference conditions (such as high-load complex spans, the Stroop task, or the anti-saccade paradigm), high-WMC individuals show sustained, focused DLPFC activation coupled with effective functional connectivity to posterior task-relevant areas. Low-WMC individuals, by contrast, exhibit erratic, inconsistent DLPFC recruitment, with their control networks frequently showing transient drops in activation that coincide directly with behavioral lapses of attention, proactive interference intrusions, and cognitive errors.

8. Administration Protocols, Standardization, and Forward vs. Backward Modalities

8.1 Standardized Administration of the Corsi Block-Tapping Paradigm

Administering the manual Corsi Block-Tapping Test with psychometric precision requires strict adherence to standardized spatial and temporal parameters. Deviations in examiner behavior can introduce substantial measurement error. Standardized clinical protocol, established through widely adopted norms (such as those by Kessels et al., 2000), dictates the following procedural conditions:

• Physical Layout and Spatial Alignment: The standardized Corsi board must be placed on a level table directly between the examiner and the participant. The baseboard is positioned so that the non-numbered face of the nine blocks faces the participant, while the numbered surface is visible exclusively to the examiner. The distance from the participant’s chest to the board should be standardized (typically 20–25 cm) to ensure comfortable manual reach across the entire spatial array without necessitating upper-body postural shifts.
• Temporal Tapping Velocity: The examiner must tap the designated sequence of blocks at an exact cadence of one block per second. The tapping action should be executed using the index finger, tapping the flat top surface of each block cleanly without sliding across the board. The examiner must avoid pausing longer between specific blocks, as rhythmic temporal grouping artificially inflates spatial span scores.
• Instructional Delivery: The participant is instructed to observe the sequence carefully, keep their hands resting flat on the table until the examiner completely finishes the tapping sequence, and then reproduce the exact sequence of blocks in the identical order.
• Discontinuation and Scoring Mechanics: Testing commences with sequences of two or three blocks, presenting two or three trials per sequence length. If the participant correctly reproduces at least one trial at a given length, the sequence is increased by one block. Testing is discontinued when the participant fails all trials at a specific sequence length. Scoring yields two primary metrics:
  1. Corsi Span: The maximum sequence length at which the participant successfully passed at least one trial.
  2. Corsi Product Score (or Total Raw Score): The total number of correct trials multiplied by the span length, or the simple sum of all correctly reproduced trials across the entire test session. The Product Score frequently exhibits superior psychometric sensitivity because it captures trial-level consistency rather than an isolated threshold.

8.2 Standardized Administration of the Digit Span Paradigm

The modern clinical standardization of the Digit Span task—exemplified in the Wechsler Adult Intelligence Scale (WAIS-IV)—comprises three distinct administrative sub-conditions: Digit Span Forward, Digit Span Backward, and Digit Span Sequencing. Each condition evaluates a distinct tier of cognitive processing.

The administrative protocol mandates strict acoustic controls. The examiner must articulate the digits in a clear, monotonic voice, maintaining a steady presentation rate of precisely one digit per second. The examiner must intentionally suppress conversational prosody, avoiding any pitch inflection until the final digit of the sequence, where a slight downward inflection signals that the sequence has terminated. Digits must not be grouped into rhythmic couplets or triplets (e.g., reciting “4-2… 9-8” must be strictly avoided; the cadence must be a uniform “4… 2… 9… 8”).

The procedural structure across the three WAIS-IV subtests is systematically organized:

• Digit Span Forward: The participant repeats the digits in the exact linear sequence presented. This subtest serves as the primary metric of auditory sensory reception, phonological storage capacity, and subvocal maintenance.
• Digit Span Backward: The participant is instructed to repeat the spoken digits in the exact reverse order. If the examiner delivers “7 – 1 – 9”, the participant must respond “9 – 1 – 7”. This test requires active sequence manipulation, attentional suppression, and executive reordering.
• Digit Span Sequencing: Introduced to provide a purer measure of mental manipulation and working memory updating, this subtest requires the participant to listen to an unordered string of digits (e.g., “3 – 8 – 1 – 6”) and repeat them back arranged in ascending numerical order (e.g., “1 – 3 – 6 – 8”). This requires the continuous updating of an abstract mental number line, imposing high executive load while minimizing the specific reverse-scanning mechanics of the backward task.

Each sequence length features two distinct trials consisting of completely different digit strings. Testing proceeds incrementally until the participant fails both trials at a given sequence length, at which point the subtest is discontinued. The examiner records the total raw score across all passed trials, as well as the maximum span length achieved for each of the three sub-modalities, allowing for clinical discrepancy analysis.

8.3 Cognitive Divergence in Backward Transformations

While clinical tradition often assumes that Backward Corsi and Backward Digit Span represent identical cognitive transformations across different sensory modalities, cognitive task analyses reveal profound structural and computational divergences between them.

In the Backward Digit Span task, the backward transformation introduces significant cognitive resistance. Spoken numbers are processed sequentially through time, creating a powerful forward temporal vector. When asked to reverse the string, the participant must actively dismantle this temporal association. They must hold the forward trace in a suspended state, suppress the urge to verbalize the initial digits, scan backward through the phonological representation to isolate the terminal item, output that item, and repeat the search for each preceding item. This requires substantial working memory capacity, executive control, and resistance to proactive interference, which is why healthy adults almost universally show a Backward Digit Span that is 1 to 2 digits shorter than their Forward Digit Span.

In striking contrast, the Backward Corsi Block-Tapping Test frequently produces a psychometric paradox: healthy participants routinely achieve Backward Corsi spans that are equal to, or occasionally even exceed, their Forward Corsi spans. This phenomenon, well-documented in the neuropsychological literature (Wilde, Wilde, & Vogel, 2008), stems from the spatial nature of the stimulus array.

When an examiner taps a sequence of blocks, the participant encodes not just isolated serial order, but an integrated, holistic geometric path across the testing board. This visuospatial representation forms a static, configural trace within the right posterior parietal cortex. Traversing a spatial path in reverse—from its destination back to its origin—is an operation for which the human spatial navigation system is naturally optimized. Retracing a path in physical space does not require destroying the original memory trace; it merely requires reading out an integrated spatial map from the opposite terminal vector. Consequently, the backward transformation in the Corsi test does not consistently impose the severe central executive bottleneck that characterizes the Backward Digit Span, making Backward Corsi a less demanding metric of executive attention than its verbal analogue.

9. Individual Differences, Cognitive Aging, and Fluid Intelligence ($G_f$)

9.1 Predicting Fluid Intelligence: The Superiority of Engle’s Construct

One of the central contributions of modern differential psychology has been resolving the relationship between Working Memory Capacity (WMC) and general fluid intelligence ($G_f$). Fluid intelligence, epitomized by non-verbal abstract reasoning batteries such as Raven’s Advanced Progressive Matrices and Cattell’s Culture Fair Test, reflects the capacity to solve novel problems, identify underlying patterns, extrapolate logical rules, and adapt to unfamiliar scenarios independently of acquired cultural knowledge.

For decades, cognitive psychometrics was puzzled by the modest relationship between simple span tasks and fluid reasoning. As summarized by Kane et al. (2004), across dozens of independent empirical datasets, the direct correlation between simple Forward Digit Span or simple Forward Corsi and fluid intelligence rarely exceeded $r = 0.20$ to $0.30$. If short-term memory capacity represented the fundamental core of human mental bandwidth, why did individuals with exceptionally high digit spans frequently demonstrate mediocre matrix reasoning scores?

Randall Engle’s executive attention theory resolved this mystery through structural equation modeling. By decomposing span performance into domain-specific storage versus domain-general executive attention, Engle revealed the underlying psychometric dynamics:

When fluid reasoning tasks like Raven’s Matrices are examined, they do not merely demand passive storage. Instead, they require the test-taker to hold several abstract rules in an active state (e.g., shape alteration, color inversion, size reduction), test potential hypotheses, update intermediate solutions, and systematically shield the current goal from competing, incorrect visual options. This exact capability—interference suppression and goal maintenance—is what complex span tasks measure. When the shared variance of executive attention is extracted from complex spans, it explains between 30% and 50% of the total variance in general fluid intelligence. Simple Digit and Corsi spans predict fluid intelligence only to the modest degree that they implicitly demand executive control; once their domain-specific storage variance is isolated, their predictive utility for $G_f$ drops to near zero.

9.2 Developmental Trajectories from Childhood to Adolescence

The developmental trajectories of the cognitive capacities measured by the Digit Span, Corsi Block-Tapping, and complex span tasks exhibit distinct maturation timelines that mirror the underlying neuroanatomical development of the human brain.

In early childhood (ages 4 to 8), simple span capacities expand rapidly. Forward Digit Span increases from an average of 2–3 items at age four to approximately 5 items by age seven. This early growth is driven primarily by the development of the articulatory rehearsal mechanism within the phonological loop. Young children under age seven rarely engage in spontaneous subvocal rehearsal; their retention reflects the raw decay time of the passive phonological store. As expressive language networks mature and inner speech develops around age seven, children begin actively cycling phonological traces, leading to an immediate increase in verbal span scores. Corsi Block performance displays a parallel trajectory, rising steadily as visual tracking networks, spatial scanning, and manual motor control become increasingly integrated across the posterior parietal cortex.

However, performance on complex span tasks that measure Engle’s executive attention construct exhibits a protracted developmental timeline. Complex working memory capacity continues to mature throughout late adolescence and into early adulthood, reaching its peak between 20 and 25 years of age. This delayed maturation corresponds directly to the prolonged structural myelination and synaptic pruning of the prefrontal cortex and the frontoparietal control network. Preadolescent children perform poorly on complex span tasks not because their phonological or spatial storage bins are inadequate, but because their immature prefrontal cortices struggle to maintain primary task goals when disrupted by interleaved processing tasks, leaving them highly vulnerable to proactive interference.

9.3 Cognitive Aging and Differential Decay Rates

The effects of normal biological aging reveal clear dissociations between spatial working memory, verbal short-term memory, and executive attention. A consistent finding in cognitive gerontology is the asymmetric decline observed between the Digit Span and Corsi Block-Tapping tests.

Auditory-verbal Forward Digit Span is remarkably resilient to the effects of normal aging. Healthy adults in their seventies and eighties frequently maintain Forward Digit Spans of 6 to 7 items, showing only slight reductions compared to their younger counterparts. The phonological loop, stabilized by decades of overlearned language processing, remains relatively robust throughout healthy senescence.

In stark contrast, spatial working memory capacity—as measured by the Corsi Block-Tapping Test—exhibits rapid, linear age-related decline. Healthy older adults demonstrate pronounced reductions in both forward and backward Corsi spans, with average performance dropping to 3 or 4 blocks by the eighth decade of life. This vulnerability reflects the preferential degradation of the right posterior parietal cortex, frontoparietal white matter tracts, and hippocampal volume during aging. Older adults exhibit greater difficulty encoding spatial coordinates and dynamic movement trajectories than repeating acoustic tokens.

Furthermore, when evaluated through Randall Engle’s executive framework, cognitive aging is heavily characterized by an inhibition deficit (Hasher & Zacks, 1988). Older adults show catastrophic performance declines on complex span tasks and high-interference simple span conditions (such as Backward Digit Span and supraspan tasks). Aging disrupts the prefrontal cortex’s ability to clear irrelevant information from the focus of attention. As a result, older adults accumulate severe proactive interference across successive testing trials, causing their working memory performance to decline under conditions requiring cognitive control and interference suppression.

10. Clinical Neuropsychology: Differential Diagnostics and Pathological Profiles

10.1 Neurodegenerative Pathologies: Alzheimer’s Disease and Frontotemporal Dementia

The comparative administration of the Digit Span task and the Corsi Block-Tapping Test provides valuable diagnostic information in the differential evaluation of neurodegenerative dementias, particularly when distinguishing between Alzheimer’s Disease (AD) and Behavioral Variant Frontotemporal Dementia (bvFTD), as well as distinct variants of Primary Progressive Aphasia (PPA).

In typical amnestic Alzheimer’s disease, neuropathology initiates within the transentorhinal and hippocampal formations before spreading to the posterior temporoparietal association cortices. Consequently, early-stage AD patients often demonstrate an asymmetric collapse in Corsi Block performance. Their immediate spatial span is significantly truncated, and their capacity to recall spatial paths over brief delays drops toward floor levels, directly reflecting the degeneration of the right hippocampal-parietal network. In contrast, Forward Digit Span can remain relatively preserved in mild AD, as the left perisylvian articulatory core is spared until later in the disease progression.

A different psychometric profile emerges across the primary progressive aphasias:

• Logopenic Variant PPA (lvPPA): Typically associated with underlying Alzheimer’s pathology affecting the left temporoparietal junction, lvPPA patients exhibit an isolated collapse of the phonological loop. Their Forward Digit Span is severely impaired (often dropping to 1 or 2 digits), while their Corsi Block performance remains entirely within normal limits, providing a clear double dissociation.
• Semantic Variant PPA (svPPA): Associated with anterior temporal lobe atrophy, svPPA patients typically present with intact Forward Digit Span and normal Corsi spans, as their primary deficit involves semantic representation rather than phonological or spatial sequence maintenance.

In Behavioral Variant Frontotemporal Dementia (bvFTD), where degeneration targets the orbitofrontal, anterior cingulate, and dorsolateral prefrontal cortices, patients present with profound deficits across complex working memory tasks and backward span transformations, as conceptualized by Engle. While their raw Forward Digit Span and Forward Corsi spans may appear preserved on initial testing, they struggle when required to suppress proactive interference, resist distraction, or perform mental manipulations, reflecting the loss of prefrontal executive control networks.

10.2 Neurodevelopmental and Psychiatric Disorders: ADHD and Schizophrenia

The application of Engle’s executive attention theory has clarified the cognitive profiles associated with common neurodevelopmental and psychiatric disorders, specifically Attention-Deficit/Hyperactivity Disorder (ADHD) and Schizophrenia.

In clinical evaluations of children and adults with ADHD, traditional testing often yielded confusing outcomes: many individuals with ADHD scored within the average range on simple Forward Digit Span and Forward Corsi tests. These findings led some clinicians to question whether working memory deficits were truly a central feature of the disorder. However, viewing ADHD through Engle’s framework clarifies the profile:

• ADHD does not represent an impairment in domain-specific phonological or spatial storage capacity. The physical limits of their storage buffers are entirely intact.
• ADHD represents an impairment in executive attention and top-down goal maintenance. When individuals with ADHD are evaluated using complex span batteries (such as O-Span or Symmetry Span), or when simple spans are administered under conditions of high proactive interference, performance drops significantly. The prefrontal frontostriatal dopamine circuits that sustain goal focus against distraction fail to maintain activation, leading to frequent attentional lapses and task errors.

In Schizophrenia, working memory dysfunction represents a central cognitive feature of the illness that strongly predicts poor functional and occupational outcomes. Neuroimaging and cognitive studies demonstrate that patients with schizophrenia exhibit profound hypofunction within the dorsolateral prefrontal cortex (hypofrontality) coupled with abnormal dopamine D1 receptor signaling. Consequently, schizophrenia patients show catastrophic deficits on tasks requiring active sequence manipulation, interference resistance, and context updating.

While a patient with schizophrenia may demonstrate a normal Forward Digit Span of 6 digits, their Backward Digit Span, Digit Sequencing, and Corsi spans show marked impairment. They struggle to maintain the mental sequence when internal cognitive resistance is introduced, as their prefrontal control networks cannot shield fragile mnemonic traces from noise and cognitive interference.

10.3 Traumatic Brain Injury (TBI) and Stroke Syndromes

In neurotrauma and vascular neurology, the Digit Span and Corsi Block-Tapping tests serve as important tools for mapping functional deficits, establishing baselines, and monitoring rehabilitative trajectories.

In Traumatic Brain Injury (TBI), particularly closed-head injuries resulting from high-velocity motor vehicle accidents or concussions, the primary biomechanical pathology is diffuse axonal injury (DAI). DAI preferentially shears and stretches long, myelinated white-matter tracts, prominently including the superior longitudinal fasciculus and frontostriatal pathways. As a consequence of this diffuse network disconnection, TBI patients exhibit pronounced slowing of information processing speed and heightened vulnerability to distraction. While their Forward Digit Span may recover to normal limits, their Backward Digit Span and complex span scores remain persistently depressed, serving as an objective marker of executive attention deficits in post-concussive syndromes.

In ischemic and hemorrhagic stroke syndromes, the two tests provide direct lateralizing evidence:

• Left middle cerebral artery (MCA) strokes involving the superior temporal gyrus, inferior parietal lobule, or inferior frontal gyrus cause severe reductions in Digit Span performance, often presenting as conduction aphasia with impaired phonological repetition.
• Right MCA strokes involving parietal-frontal networks produce profound spatial sequence deficits on the Corsi Block-Tapping Test. Clinicians utilize the Corsi test to identify spatial sequencing apraxia and to track the recovery of non-verbal cognitive processing during stroke rehabilitation.

Additionally, focal lesions to the cerebellum produce surprising secondary impacts on both span paradigms. Historically viewed exclusively as a motor structure, the cerebellum is now recognized as an important cognitive modulator via extensive cerebello-thalamo-cortical loops. Right cerebellar lesions impair verbal digit span by disrupting the smooth internal timing of subvocal articulatory rehearsal, while left cerebellar lesions impair Corsi block performance by degrading the motor timing required for internal spatial trajectory planning.

11. Digital Innovations, Modern Psychometrics, and Computational Modeling

11.1 Computerized Corsi Implementations (e.g., e-Corsi, PEBL)

The migration of the Corsi Block-Tapping Test from a manual wooden apparatus to computerized platforms—such as the e-Corsi, the Psychology Experiment Building Language (PEBL), and mobile touchscreen applications—has resolved long-standing psychometric limitations while introducing new analytical metrics.

Manual administration of the Corsi test was inherently vulnerable to examiner bias: variations in finger-travel velocity, minor hesitations between targets, and unintentional ocular cueing introduced uncontrolled variance. Computerized paradigms eliminate these confounds. In a computerized setup, the standardized nine-block asymmetrical array is rendered on a high-resolution touchscreen display. Blocks illuminate sequentially using high-contrast color shifts (e.g., shifting from blue to yellow) for an exact duration of 1,000 milliseconds per block, separated by a precise 1,000-millisecond inter-stimulus interval.

Beyond standardizing presentation pacing, computerized testing unlocks rich, dynamic behavioral metrics that were impossible to capture via manual observation:

• Inter-Tap Latency Analysis: Instead of simply scoring a trial as correct or incorrect, digital software records the exact millisecond-level latency between successive block touches. Prolonged latencies at specific sequence transitions reveal the internal boundaries of mental chunking strategies.
• Spatial Trajectory Deviation: Computerized systems can track the physical path of the participant’s finger across the screen, calculating motor trajectory angles, spatial hesitation loops, and directional corrections.
• Spatial Proximity Error Scoring: In traditional scoring, tapping an incorrect block was a binary failure. Digital algorithms can quantify the metric spatial distance between the target block and the erroneously tapped block. Tapping an immediately adjacent block reflects a minor metric encoding error, whereas tapping a block in the opposite quadrant indicates a complete breakdown of spatial representation.

11.2 Automated Complex Span Testing: Engle’s Open-Access Methodologies

In parallel with the digitization of spatial testing, Randall Engle’s laboratory revolutionized experimental psychometrics by creating and freely distributing automated, standardized versions of their complex span batteries. Early implementations of the Operation Span (O-Span) and Reading Span (R-Span) required direct, one-on-one manual administration with index cards or early software, making large-scale testing labor-intensive and susceptible to administrator variations in reading pace and scoring.

To overcome these methodological bottlenecks, Unsworth, Heitz, Schrock, and Engle (2005) developed the Automated Operation Span (A-Span), followed by automated versions of the Symmetry Span and Reading Span. These open-access tools, built on standardized testing engines, automated the entire presentation, timing, and scoring architecture:

• Dynamic Time-Out Calibration: To prevent participants from trading speed for accuracy, the automated software establishes an individualized baseline for processing speed. In the initial phase, participants solve simple mathematical problems as quickly as possible. The software calculates the participant’s mean response latency plus 2.5 standard deviations, establishing a strict, personalized deadline for the processing component during subsequent dual-task testing.
• Standardized Scoring Algorithms: The software computes two psychometrically validated metrics:
  1. Partial-Credit Unit (PCU) Score: Calculates the proportion of correctly recalled elements within each sequence, regardless of whether the entire sequence was completely correct. PCU scoring maximizes psychometric sensitivity and exhibits superior internal consistency ($r > 0.80$).
  2. Absolute Score: Reflects the sum of all elements recalled in entirely error-free trials, capturing high-threshold performance.
• Strict Processing Accuracy Filtering: To prevent participants from strategically ignoring the processing task (e.g., guessing math equations to focus entirely on memorizing the letters), the software tracks overall processing accuracy. Data from participants whose processing accuracy falls below an 85% threshold are automatically flagged or excluded, ensuring the construct validity of the executive attention measurement.

11.3 Computational and Mathematical Models of Serial Recall

The empirical data generated by the Digit Span, Corsi Block-Tapping, and complex span paradigms have provided empirical constraints for computational and mathematical models of human serial recall. How does the human brain encode the temporal order of events?

In the domain of verbal memory, computational architectures have largely moved away from simple associative chaining models (which proposed that Item 1 triggers Item 2, which triggers Item 3). Chaining models fail because an error on Item 2 should theoretically cause the immediate collapse of the entire subsequent sequence, an outcome contradicted by empirical data. Instead, modern computational models—such as the OSCAR (Oscillator-Based Serial Order Recall) model (Brown, Preece, & Hulme, 2000) and the Burgess & Hitch Phonological Loop Model—rely on positional coding. Items are bound to a dynamic, internal multi-frequency neural clock or context vector. Retrieval involves resetting this internal clock and projecting the context vector forward, sequentially activating the items bound to each successive temporal coordinate.

In the spatial domain, the Corsi Block-Tapping Test has been modeled computationally using Continuous Attractor Neural Networks (CANN). In a CANN model, each block location corresponds to a stable bump of localized neuronal excitation within a continuous two-dimensional network representing physical space. Presenting a sequence of taps establishes a sequence of localized attractor states. Serial order is maintained through a process of synaptic gain modulation or decaying activation gradients: the first block tapped receives the highest activation baseline, and each subsequent block receives a progressively lower activation baseline. During recall, a winner-take-all competitive inhibition network reads out the attractor bumps in order of descending activation strength.

From the perspective of Randall Engle’s executive attention theory, computational modeling formalizes executive control as dynamic synaptic gain control. Within these models, the prefrontal cortex does not store the specific items; instead, it injects top-down excitatory currents into task-relevant neural populations in posterior cortices, preventing decaying attractor bumps from dissolving into the background noise of proactive interference.

12. Theoretical Synthesis: Reconciling Domain-Specific Buffers with Executive Attention

12.1 A Hierarchical Architecture of Human Working Memory

After five decades of empirical research, clinical debate, and theoretical modeling, the cognitive sciences have achieved a nuanced reconciliation between Alan Baddeley’s structural multicomponent model and Randall Engle’s executive attention framework. Rather than viewing these theories as mutually exclusive, modern cognitive psychology integrates them into a unified, hierarchical architecture of human working memory.

This hierarchical framework, which draws from the embedded-process models of Nelson Cowan (1999) and Klaus Oberauer (2002) alongside Engle’s and Baddeley’s formulations, organizes immediate memory across three functional tiers:

1. The Activated Long-Term Memory (LTM) Base: The foundational substrate of all cognitive representations—whether phonological forms, semantic definitions, or spatial coordinate maps—consists of the vast network of long-term memory. Sensory stimuli automatically activate these representations, elevating them above a resting baseline.
2. Domain-Specific Temporary Buffers / Slave Systems: Baddeley’s phonological loop and visuospatial sketchpad operate as specialized, localized sensorimotor maintenance networks. The phonological store and articulatory loop (indexed by the Digit Span task) and the visual cache and inner scribe (indexed by the Corsi Block-Tapping Test) provide domain-specific mechanisms to preserve activation over brief delays via automated sensorimotor loops without demanding heavy prefrontal executive resources.
3. The Central Executive / Focus of Attention: At the top of the hierarchy sits Engle’s domain-general executive attention system. Mediated by the prefrontal cortex and the frontoparietal control network, this supervisory system manages the focus of attention. It maintains goal-relevant task sets, directs attention across posterior representations, coordinates complex mental manipulations, and provides the top-down inhibitory control needed to shield fragile traces from internal and external interference.

Within this hierarchical synthesis, the Corsi Block-Tapping Test and the Digit Span task are recognized as tests that tap both Tier 2 and Tier 3 mechanisms. In their simple, forward iterations under low interference, they load primarily on Tier 2 domain-specific maintenance. But as sequence lengths increase, backward manipulations are introduced, or cognitive resistance is applied, they increasingly demand Tier 3 executive attention resources, engaging the domain-general prefrontal mechanisms characterized by Randall Engle.

12.2 The Role of Interference Suppression across Verbal and Spatial Modalities

A central insight gained from synthesizing Engle’s work with classical span testing is that interference suppression represents the common computational mechanism underlying human immediate memory limitations across both verbal and spatial modalities.

Historically, capacity limits were viewed as static physical boundaries: the phonological loop ran out of temporal recording tape after two seconds, or the visuospatial sketchpad ran out of discrete coordinate slots after five blocks. Modern evidence demonstrates that this container-based view is largely incomplete. Immediate memory capacity is fundamentally limited by the rapid accumulation of cue overload and proactive interference.

When an individual takes a Digit Span test, the primary challenge across repeated trials is not simply hearing the numbers; it is resolving the competition between the numbers heard on the current trial and the echoes of numbers heard on the preceding five trials. In the Corsi Block-Tapping Test, the participant does not view the nine blocks on a clean slate; by Trial 6, virtually every block on the board has been tapped in various sequence configurations, creating a dense field of competing spatiotemporal vectors. Performance fails when the retrieval cues become overloaded with conflicting memories.

Executive attention, as conceptualized by Engle, is the mechanism that counters this informational interference. Whether suppressing phonological competitors in the supramarginal gyrus or resolving overlapping spatial trajectories in the intraparietal sulcus, the prefrontal cortex uses the identical operational algorithm: it enhances the gain on task-relevant target representations and suppresses irrelevant, non-target noise. Thus, when an individual demonstrates a superior working memory capacity across any modality, they are demonstrating superior attentional control over cognitive interference.

12.3 Directions for Future Empirical and Neurocognitive Inquiry

As the study of cognitive span paradigms moves forward, several empirical frontiers promise to refine our understanding of human working memory:

• High-Density Intracranial Electrophysiology (iEEG): While fMRI has localized the macroscopic circuits of Digit Span and Corsi performance, it lacks the millisecond temporal resolution needed to observe real-world neural dynamics. Intracranial electrophysiology in surgical epilepsy patients allows researchers to record direct local field potentials across the prefrontal cortex, parietal cortex, and hippocampus simultaneously during span execution. These studies are unraveling how phase-amplitude coupling between low-frequency theta oscillations (4–8 Hz) and high-frequency gamma bursts (30–80 Hz) establishes a discrete multiplexing code that preserves serial order across both phonological and spatial domains.
• Machine Learning and Dynamic Behavioral Tracking: Applying deep neural networks and computer-vision algorithms to computerized span tasks allows for the real-time decoding of a participant’s cognitive strategies. By analyzing eye movements, inter-tap latency patterns, and micro-hesitations during digital Corsi or Digit Span execution, machine learning models can detect whether a participant is shifting between spatial chunking, verbal recoding, or executive goal loss, enabling adaptive, individualized cognitive assessment.
• Targeted Neuromodulation and Cognitive Therapeutics: Understanding the frontoparietal networks that govern executive attention opens new avenues for therapeutic intervention. Transcranial magnetic stimulation (TMS) and high-definition transcranial direct current stimulation (HD-tDCS) targeting the dorsolateral prefrontal cortex are being evaluated for their potential to enhance interference resistance in individuals suffering from neurotrauma, ADHD, schizophrenia, and age-related cognitive decline, translating Engle’s theoretical frameworks into clinical interventions.

Conclusion

The journey from the wooden blocks of Philip Corsi’s laboratory to the controlled experimental batteries of Randall Engle illustrates the evolution of modern cognitive psychology. For over a century, simple span paradigms like the Digit Span and Corsi Block-Tapping tests served as foundational instruments in neuropsychological assessment, establishing double dissociations between left-hemisphere auditory-verbal processing and right-hemisphere visuospatial tracking. These paradigms mapped the functional architecture of domain-specific buffers, demonstrating how the human brain maintains immediate sensory representations over brief temporal intervals.

Yet, it was the theoretical intervention of Randall Engle that fundamentally redefined our understanding of these instruments. By demonstrating that simple storage capacity explains only a fraction of complex intellectual behavior, Engle shifted the empirical paradigm from passive storage containers to dynamic executive attention. His framework revealed that working memory capacity is ultimately an index of attentional control: the capacity to maintain goal-relevant representations in an active state against proactive interference, distraction, and internal cognitive competition. Today, whether evaluated via manual standardized tests, computerized touchscreens, or intracranial electrophysiology, the Digit Span and Corsi Block-Tapping paradigms remain indispensable. Viewed through the lens of executive attention, they provide clear windows into the multi-layered cognitive and neural architecture that enables the human mind to coordinate thought, action, and reasoning.

References

• Atkinson, R. C., & Shiffrin, R. M. (1968). Human memory: A proposed system and its control processes. In K. W. Spence & J. T. Spence (Eds.), The Psychology of Learning and Motivation (Vol. 2, pp. 89–195). Academic Press. https://doi.org/10.1016/S0079-7421(08)60422-3
• Baddeley, A. D., & Hitch, G. (1974). Working memory. In G. H. Bower (Ed.), The Psychology of Learning and Motivation (Vol. 8, pp. 47–89). Academic Press. https://doi.org/10.1016/S0079-7421(08)60452-1
• Brown, G. D., Preece, T., & Hulme, C. (2000). Oscillator-based memory for serial order. Psychological Review, 107(1), 127–181. https://doi.org/10.1037/0033-295X.107.1.127
• Burgess, G. C., Gray, J. R., Conway, A. R., & Braver, T. S. (2011). Neural correlates of working memory capacity: The impact of factor structure. NeuroImage, 56(3), 1806–1816. https://doi.org/10.1016/j.neuroimage.2011.02.062
• Conway, A. R., Cowan, N., & Engle, R. W. (2001). The cocktail party phenomenon revisited: The importance of working memory capacity. Psychonomic Bulletin & Review, 8(2), 331–335. https://doi.org/10.3758/BF03196169
• Corsi, P. M. (1972). Human memory and the medial temporal region of the brain (Doctoral dissertation, McGill University). McGill University Theses and Dissertations.
• Cowan, N. (1999). An embedded-processes model of working memory. In A. Miyake & P. Shah (Eds.), Models of Working Memory: Mechanisms of Active Maintenance and Executive Control (pp. 62–101). Cambridge University Press. https://doi.org/10.1017/CBO9781139174909.006
• Daneman, M., & Carpenter, P. A. (1980). Individual differences in working memory and reading. Journal of Verbal Learning and Verbal Behavior, 19(4), 450–466. https://doi.org/10.1016/S0022-5371(80)90312-6
• Engle, R. W. (2002). Working memory capacity as executive attention. Current Directions in Psychological Science, 11(1), 19–23. https://doi.org/10.1111/1467-8721.00160
• Engle, R. W., Kane, M. J., & Tuholski, S. W. (1999). Individual differences in working memory capacity and what they tell us about controlled attention, general fluid intelligence, and functions of the prefrontal cortex. In A. Miyake & P. Shah (Eds.), Models of Working Memory: Mechanisms of Active Maintenance and Executive Control (pp. 102–134). Cambridge University Press. https://doi.org/10.1017/CBO9781139174909.007
• Engle, R. W., Tuholski, S. W., Laughlin, J. E., & Conway, A. R. (1999). Working memory, short-term memory, and general fluid intelligence: A latent-variable approach. Journal of Experimental Psychology: General, 128(3), 309–331. https://doi.org/10.1037/0096-3445.128.3.309
• Hasher, L., & Zacks, R. T. (1988). Working memory, comprehension, and aging: A review and a new view. In G. H. Bower (Ed.), The Psychology of Learning and Motivation (Vol. 22, pp. 193–225). Academic Press. https://doi.org/10.1016/S0079-7421(08)60047-X
• Jacobs, J. (1887). Experiments on “prehension”. Mind, 12(45), 75–79. https://doi.org/10.1093/mind/os-XII.45.75
• Kane, M. J., & Engle, R. W. (2000). Working-memory capacity, proactive interference, and divided attention: Limits on long-term memory retrieval. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26(2), 336–358. https://doi.org/10.1037/0278-7393.26.2.336
• Kane, M. J., & Engle, R. W. (2002). The role of prefrontal cortex in working-memory capacity, executive attention, and general fluid intelligence: An individual-differences perspective. Psychonomic Bulletin & Review, 9(4), 637–671. https://doi.org/10.3758/BF03196323
• Kane, M. J., Hambrick, D. Z., Tuholski, S. W., Wilhelm, O., Payne, T. W., & Engle, R. W. (2004). The generality of working-memory capacity: A latent-variable approach to verbal and visuospatial memory with and without executive attention. Journal of Experimental Psychology: General, 133(2), 189–217. https://doi.org/10.1037/0096-3445.133.2.189
• Kessels, R. P., van Zandvoort, M. J., Postma, A., Kappelle, L. J., & de Haan, E. H. (2000). The Corsi Block-Tapping Task: Standardization and normative data. The Clinical Neuropsychologist, 14(2), 252–258. https://doi.org/10.1076/1385-4046(200005)14:2;1-Z;FT252
• Logie, R. H. (1995). Visuo-spatial working memory. Lawrence Erlbaum Associates. https://doi.org/10.1093/acprof:oso/9780198523093.001.0001
• 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
• Milner, B. (1971). Interhemispheric differences in the localization of psychological processes in man. British Medical Bulletin, 27(3), 272–277. https://doi.org/10.1093/oxfordjournals.bmb.a070866
• Oberauer, K. (2002). Access to information in working memory: An inspection of the executive attention model. Journal of Experimental Psychology: Learning, Memory, and Cognition, 28(3), 411–423. https://doi.org/10.1037/0278-7393.28.3.411
• Unsworth, N., Heitz, R. P., Schrock, J. C., & Engle, R. W. (2005). An automated version of the operation span task. Behavior Research Methods, 37(3), 498–505. https://doi.org/10.3758/BF03193766
• Wechsler, D. (2008). Wechsler Adult Intelligence Scale–Fourth Edition (WAIS-IV). NCS Pearson.
• Wilde, N. J., Wilde, M. C., & Vogel, K. A. (2008). Forward and backward digit and visual span performance in an adult clinical population. The Clinical Neuropsychologist, 22(4), 633–643. https://doi.org/10.1080/13854040701416410