sWorking Memory Processes & RDoC Sub-construct

  • Working Memory: Refers to the ability to hold and manipulate information over short periods of time, critical for problem-solving and reasoning tasks.

  • RDoC Sub-constructs: This framework includes various dimensions such as cognitive systems, social processes, and negative valence systems, which can all intersect with working memory functions. Moreover, understanding the interactions between these constructs can provide insights into how deficits in working memory might contribute to various psychological disorders. In particular, cognitive control plays a pivotal role in overseeing working memory tasks, enabling individuals to prioritize information and suppress distractions during cognitive processing.

Key Experimental Paradigms Testing WM Assumptions

  • Complex Span Tasks (Daneman & Carpenter, 1980): Dual demand of storage (final word recall) + processing (sentence plausibility).

  • Letter–Number Sequencing (Wechsler, 1997): Hear mixed strings, repeat numbers in ascending then letters alphabetically – taps verbal + visuospatial WM.

  • N-back Task (Kirchner, 1958; popularised by Jaeggi et al., 2010): Respond when current stimulus matches NN positions back; load-dependent. 24-study fMRI meta-analysis (Owen et al., 2005) shows activation of DLPFC, VLPFC, dorsal cingulate, medial premotor, frontal poles; distinctions by content (verbal vs. spatial) & monitoring type (identity vs. location).

Embedded-Process Model (Cowan, 1999–2005)

  • WM = temporarily activated LTM; a subset enters focus of attention, directed by the Central Executive.

  • Two attentional parameters:

    • Scope (dynamic capacity; how many items).

    • Control (maintenance; resisting distraction).

  • Capacity limits stem from both temporal decay (need refreshing) and item number constraints (approx. 44 chunks; “magical four”).

Unsettled Theoretical Issues

  • Does Baddeley’s CE fully capture James’ “conscious present”?

  • Debate: Baddeley (multi-store buffers) vs. Cowan (activation-based); both agree on limited capacity & role of attention but differ in metaphor (homuncular vs. neurodynamic).

  • Baddeley (2012) asserts attention & inhibition are implicit in CE; Cowan argues specialised buffers unnecessary.

  • Consensus: neither model yet complete.

Chunking & Capacity Limits

  • Professional mnemonist (“Lightning Calculator”) could visualise 5 × 5 digit matrix; reported seeing 6 figures clearly – aligns with Cowan (activation limit ≈6\approx 6) and Miller (7 ± 27\ \pm\ 2) / Cowan 4-chunk limit.

  • Functional reasons for 3–53–5 item cap (Ma, Husain & Bays, 2014):

    • Strengths: promotes organisation (beginning–middle–end), keeps representations distinct, facilitates relational binding.

    • Weaknesses: high metabolic cost; neural firing cycles limited; inferior parietal lobe shows item-specific capacity ceiling.

Time-Based & Attention-Based Accounts

  • Engle (2002): WM capacity = ability to maintain attention & protect activation from interference rather than sheer storage.

  • Wilhelm et al. (2013) three contributors:

    1. Attentional control (filter/attenuation).

    2. Maintenance mechanisms (chunking, rehearsal, articulatory loop).

    3. Inhibition of off-target activations (precision / Bayesian updating).

  • Time-sharing: switching between processing, updating, and refreshing is limited by neurophysiological speed (synaptic, cellular, connectivity).

Working Memory vs. Executive Function (EF)

  • McCabe et al. (2010); Miyake et al. (2000): Structural-equation models show:

    • WM latent factor = temporary storage + manipulation.

    • EF factor = goal-directed behaviour (set-shifting, monitoring, updating, inhibition).

  • Terminology dispute:

    • Baddeley: EF ⊂ WM (part of CE).

    • Cowan: WM ⊂ EF.

    • Others: independent constructs.

  • Damage to prefrontal cortex (PFC) impairs all enumerated CE/EF operations (Grafman, 2002).

Open Questions & RDoC Motivation

  • Not all sensory input reaches LTM ⇒ need for filtering mechanisms pre-encoding.

  • WM must integrate: selective attention → transient echoic/iconic stores → CE-mediated manipulation; plus access to recently activated LTM.

  • National Institute of Mental Health RDoC framework reframes WM research into four measurable sub-constructs: active maintenance, flexible updating, limited capacity, interference control – relevant for psychopathology.

RDoC Sub-Constructs & Definitions

  • Active Maintenance: Holding info short-term to manipulate it (Miyake et al., 1999). Neurobasis: persistent fronto-parietal activity, synchronous oscillations, synaptic potentiation (Nyberg & Eriksson, 2016).

  • Flexible Updating: Manipulating current content based on new inputs; relies on dorsal PFC & fronto-striatal “gating” (Nyberg & Eriksson, 2016).

  • Limited Capacity: Inter-individual variation & deficits traceable to basal-ganglia integrity & dopaminergic modulation (Nyberg & Eriksson, 2016).

  • Interference Control: Achieving goals by shielding representations from distraction; contexts coded in rostrofrontal cortex; networks linked via reinforcement-learning (Nyberg & Eriksson, 2016).

Neuroscience Evidence for Each Sub-Construct

Active Maintenance

  • Fuster (1973): Monkey PFC neurons fire during delay in matching tasks.

  • Goldman-Rakic: Sustained DLPFC, parietal, striatal circuits via glutamatergic NMDA; modulated by GABA interneurons.

Flexible Updating

  • Karl Friston – Active Inference: WM as Bayesian evidence accumulation; eye saccades = hypothesis testing (Parr & Friston, 2017).

  • Basal ganglia dopamine-driven salience gating updates cortical predictions under uncertainty.

Limited Capacity

  • Capacity limit 3–53–5 chunks (Cowan, 2010).

  • Inferior parietal lobe shows item-specific neural saturation.

  • Limited cycle time of recurrent firing imposes biological ceiling; chunking/rehearsal offsets but doesn’t eliminate ceiling.

Interference Control

  • Stroop Test: Manipulate color-word congruence; load varies with salience (McLeod & MacDonald, 2000).

  • Neural network: ACC & DLPFC (conflict detection & resolution) plus parietal regions; maturation continues through adolescence.

Integrative View – Action–Perception Cycle

  • Nyberg & Eriksson (2016); Fuster (2013) diagram: perception → encoding into WM → manipulation (yellow) → action planning (orange) ↔ motivational drives (red) ↔ LTM (blue).

  • Highlights bidirectional interaction: WM both informs and is updated by sensory input and executed action.

Application to Mental Disorders (Student-Led Task)

  • Examine over- vs. under-activation of WM networks in:

    • Eating disorders (AN, BN, BED).

    • Addictions (substance, behavioural).

    • ADHD.

    • Anxiety / Depression.

    • Schizophrenia.

    • Eating Disorders: Involves disrupted patterns of cognition and behavior that can impair working memory.

    • Obsessive-Compulsive Disorder (OCD): Characterized by intrusive thoughts and compulsive behaviors, which can significantly hinder the ability to hold and manipulate information effectively.

    • Autism Spectrum Disorder (ASD): Often associated with challenges in executive functioning and social cognition, leading to difficulties in working memory that affect daily functioning.

    • Attention-Deficit/Hyperactivity Disorder (ADHD):marked by symptoms such as inattention and impulsivity, which can severely impact working memory capacity and utilization.

    • Major Depressive Disorder: This condition often leads to cognitive deficits, including impaired working memory, as individuals may struggle to maintain focus and process information due to persistent feelings of sadness and hopelessness. Schizophrenia: A severe mental disorder that can disrupt cognitive processing, resulting in significant deficits in working memory, affecting the ability to organize thoughts and manage daily tasks.

    • Bipolar Disorder: This mood disorder can cause fluctuations in cognitive function and attention, often leading to impairments in working memory, particularly during manic or depressive episodes.

Summary Bullet Recap

  • WM theories evolved from linear STM stores to multicomponent & activation-based models.

  • Core consensus: WM is capacity-limited, attention-dependent, frontal–parietal–striatal network.

  • Baddeley’s WMM emphasises specialised buffers; Cowan’s EPM treats WM as activated LTM within attentional focus.

  • RDoC crystallises research into four tractable neural processes: maintenance, updating, capacity, interference control – each with specific circuitry & clinical relevance.

  • Debates ongoing, but integrative evidence suggests WM is a dynamic interplay of sustained neural firing, synaptic plasticity, oscillatory synchrony, and dopaminergic gating constrained by metabolic cost.

  • This multifaceted nature highlights the potential for targeting specific processes in clinical interventions, offering pathways to enhance cognitive function and address deficits in various psychiatric disorders.