The Mapping Mechanism Between
Human Memory Weights and Decision Weights

Abstract

This paper proposes a unifying hypothesis: within the observable conventional cognitive channel, human decision weights are largely modulated by the memory weight table — memory retrieval fluency, emotional tags, and past outcomes collectively constitute the primary inputs to choice preferences, though decision weights are also subject to conditional modulation by current goals, physiological state, executive control, and environmental constraints. By integrating Damasio’s somatic marker hypothesis (acknowledging competing explanations), Kahneman’s dual-system theory, and Shohamy’s episodic memory–decision framework, this paper reinterprets all three as outputs at different levels of the same memory–valuation chain. vmPFC lesions provide relatively strong evidence of impaired emotional valuation interfaces; HSAM serves only as a suggestive boundary case and cannot demonstrate that “lossless global memory leads to decision paralysis.” This paper proposes a conditional mapping model D(t) = f(M, R, B, G, C, E) and a triple-layer decision architecture (the dynamic coupling of a physiological homeostasis layer, a stress layer, and a deliberative behavioral decision layer), as well as speculative application insights from this framework for personalized AI alignment. The validity of this framework is strictly delimited to the observable memory–decision channel.

The Question: Where Do Decision Weights Come From?

Humans make thousands of decisions every day — from “what to eat for breakfast” to “whether to accept this job offer.” What is the nature of these decisions? Traditional economics assumes humans are “rational agents” — maximizing utility through cost–benefit analysis of all options. But decades of behavioral economics and neuroeconomics research have thoroughly refuted this assumption.

Kahneman and Tversky’s prospect theory revealed that humans systematically deviate from rational expectations. Damasio’s studies of vmPFC lesion patients demonstrated that people who lose emotional processing capability suffer severely impaired decision-making in complex, uncertain situations. Shohamy’s episodic memory–decision research found that people prefer options that can be more successfully retrieved from memory.

These findings converge on a core question: If decision-making is not the product of pure rational computation, where do the “weights” that drive decisions actually come from?

This paper’s hypothesis is: within the conventional cognitive channel, decision weights are largely modulated by memory weights. Memory weights and decision weights are not two entirely independent systems — the former is the primary input source for the latter, although decision weights are also subject to conditional modulation by current goals, physiological state, executive control, and environmental constraints.

Mechanisms of Memory Weight Inscription

2.1 The Multidimensional Tagging System at the Encoding Stage

Memories are not treated equally at the encoding stage. The brain assigns “importance weights” to each new memory through a multidimensional tagging system. Recent EEG studies suggest that θ activity (3–8 Hz) during learning is associated with subsequent memory performance, and that slow oscillation–spindle coupling during sleep is also associated with consolidation; some studies are attempting to connect encoding-stage θ tags with sleep-stage coupling metrics. However, the complete causal chain “encoding-stage θ tag → sleep coupling density → consolidation resource allocation” still requires direct experimental verification. [Evidence level: moderate-strong at individual nodes, complete chain awaiting validation]

2.2 Reward Prediction Error: Surprise Determines Weight

The “surprise weight” dimension deserves particular attention. Research has found that the degree of memory encoding enhancement is proportional to the reward prediction error experienced — not to total reward. This means it is not “good outcomes are remembered well” but rather “unexpected outcomes are remembered well.” A single unanticipated failure can more profoundly alter the structure of the memory weight table than ten expected successes.

2.3 Graded Prioritization: Weight Propagation Through Temporal Proximity

Rewards selectively enhance the memory of event sequences leading to the reward, with the degree of enhancement as a function of proximity to the reward. More importantly, salient experiences open a temporal window that enhances the encoding of otherwise mundane memories both before and after the critical event — exhibiting graded prioritization. An emotional event is like a bomb — it not only illuminates itself but also “lights up” temporally adjacent ordinary memories while dimming unrelated ones.

2.4 Sleep Triage: Weight-Driven Selective Consolidation

The weight tags inscribed during encoding are read and executed during sleep. Sleep does not uniformly strengthen all encoded material; rather, it preferentially consolidates memories consistent with specific motivational or cognitive cues present during encoding — such as reward value, future utility, or emotional salience. Emotionally salient stimuli tend to be preferentially consolidated at the cost of neutral stimuli being sacrificed.

When multiple salience cues compete for consolidation resources, top-down directives (goal weight) can override emotional salience (emotional weight) — this provides the cognitive control capacity to “actively choose what to remember,” but its effectiveness is constrained by the intensity of emotional arousal. Extreme emotional experiences (such as trauma) are nearly impervious to top-down regulatory control.

Forgetting: The Clearing Mechanism of Weight Management

3.1 Three Mechanisms of Active Forgetting

Forgetting is not storage failure — it is the other half of the memory system. Recent research has revealed three major forgetting mechanisms:

3.2 The Functional Contributions of Forgetting to Decision-Making

Forgetting makes five layers of functional contributions:

Energy conservation — maintenance of each synapse requires a continuous ATP supply; retaining all synapses would cause the energy budget to explode. Retrieval efficiency — once low-weight information is cleared, retrieval speed for high-weight memories increases significantly. Signal-to-noise ratio optimization — with low-weight data cleared, high-weight signals become more prominent during decision-making. Conceptual abstraction — once specific details are forgotten, abstract patterns and regularities become easier to extract. Decision acceleration — fewer candidate memories competing in the ranking process means faster decision output.

From Memory Weights to Decision Weights: Unification of Three Theories

4.1 Damasio: Somatic Markers as the Physiological Expression of Weight Tags

Antonio Damasio’s (1994) somatic marker hypothesis holds that emotional processes guide decision-making behavior through bodily feelings. Within this framework, we redefine somatic markers as: the physiological expression that occurs when the emotional tags of high-weight memories are reactivated in a decision context.

When facing a choice, the brain searches for high-weight memories relevant to the current situation. These memories were tagged with emotional labels during encoding (via the catecholamine marking system). During retrieval, these tags trigger somatic responses — accelerated heartbeat, stomach contractions, muscle tension, changes in skin conductance. These bodily signals feed back to the decision system as “weight readouts,” biasing choice preferences.

4.2 Kahneman: System 1 as the Retrieval Output of Memory Weights

Daniel Kahneman’s (2011) dual-system theory distinguishes between System 1 (fast, automatic, intuitive) and System 2 (slow, effortful, logical). This framework provides the neurobiological implementation mechanism for System 1:

What this framework explains is the portion of System 1’s intuitive output that is driven by past experience, emotional tags, and memory retrieval. System 1 also encompasses perceptual processing, linguistic association, habitual actions, pattern recognition, and evolutionarily preset responses — not all of which need to be explained through emotional tags or autobiographical memory weights. Within the scope covered by this framework: high-weight memories are preferentially retrieved, and their attached emotional tags bias choice preferences — this constitutes the neurobiological implementation of the “experience-based judgment” component of System 1 intuition. [Evidence level: theoretical reinterpretation, covering part but not all of System 1]

System 2 = the prefrontal cortex’s secondary evaluation and potential override of System 1 output. It is “slow” because it requires the mobilization of working memory, logical reasoning, inhibition of System 1’s automatic output, and generation of alternatives. System 2 does not compute from scratch — it reviews and corrects System 1’s output.

4.3 Shohamy: Episodic Memory Directly Drives Choice Preferences

Daphna Shohamy (Columbia University) proposed a framework that directly connects episodic memory to value-based decision-making. Her core finding is that people prefer options that can be more successfully retrieved from memory — memory recall failure biases neural valuation processes.

The interpretation within this framework: retrieval fluency can significantly modulate valuation and choice preferences. An option that is tagged with high weight in the memory store → is more easily retrieved → the fluency of successful retrieval biases the valuation of that option → choice preference shifts toward that option. In certain tasks, this retrieval advantage approximately translates into a decision advantage — but this mapping is modulated by task goals, external rewards, current state, and executive control, and is not an unconditional one-to-one mapping. [Evidence level: moderate-strong, conditionally valid]

Unified Architecture of Three Theories:

Memory Weight Table (built through encoding + triage + forgetting)
      ↓
Decision context triggers → retrieval of relevant high-weight memories
      ↓
Damasio: Emotional tags → somatic response → bias choice
Kahneman: High-weight priority output → System 1 "intuition"
Shohamy: Retrieval fluency → valuation preference
      ↓
All three describe different facets of the same process
      ↓
Final output: Behavioral decision

4.4 The Conditional Mapping Model

To avoid the over-reduction of “decision weight = memory weight,” this paper proposes the following conditional mapping model:

Decision Weight D(t) = f(M, R, B, G, C, E)

Where:
  M = Memory weight (cumulative result of encoding weight + consolidation weight)
  R = Match between current retrieval cues and the memory store
  B = Current physiological state (hunger/fatigue/pain and other homeostatic parameters)
  G = Current goals and task demands
  C = Cognitive control / metacognition (prefrontal capacity to review and override automatic output)
  E = External environmental constraints (social norms, immediate incentives, physical limitations)

The core claim of this paper: In most everyday situations, M is the dominant input to D(t) —
but D(t) is not a direct copy of M; rather, it is the conditional output of M
under the joint modulation of R, B, G, C, and E.
When B is extreme (e.g., extreme hunger), B can override M;
when C intervenes forcefully (e.g., deliberate reflection), C can correct M's automatic output.

This model preserves the core thesis that “memory weight is the dominant input to decision-making” while avoiding the over-reduction of “decisions are entirely determined by memory.” Below are the qualitative interaction rules among the six variables and typical scenario predictions:

Triple-Layer Decision Architecture: Physiological Drive, Stress Interrupt, and Memory Sorting

Human behavior is not entirely driven by memory weights. In fact, by behavioral volume, the vast majority of human “decisions” each day do not pass through the memory system at all — they are direct outputs of the biological-physiological system. This paper proposes a triple-layer decision architecture, from the lowest layer of physiological drive to the highest layer of deliberative behavioral decision-making, with strict priority relationships between each layer.

5.1 Bottom Layer: The Physiological Homeostatic Decision System — High-Frequency Automatic Drive

Eating when hungry, drinking when thirsty, sleeping when fatigued, seeking warmth when cold, avoiding stimuli when in pain, voiding when the bladder is full — the fundamental driving forces behind these behaviors originate from the hypothalamus, brainstem, and autonomic nervous system, operating through homeostatic feedback loops. At the most basic level, they do not rely on emotional tag comparison or deliberative participation from the prefrontal cortex.

The primary weight signals for these physiological decisions come from homeostatic parameters: blood glucose levels, osmolarity, body temperature, hormone concentrations, nociceptive signals, and so on. But it must be noted that the physiological and memory layers are not absolutely isolated — extensive bidirectional interplay exists between them. Classical conditioning research has demonstrated that merely recalling a restaurant memory can trigger ghrelin secretion and hunger; traumatic memory flashbacks can directly activate stress physiological responses (accelerated heartbeat, sweating); and even pain anticipation (based on memory) can alter actual pain perception. Therefore, a more accurate description is: the physiological layer has its own independent weight system (homeostatic parameters), but this system is continuously modulated and biased by the memory layer. The relationship between the two is not “absolutely isolated hierarchical suppression” but a coupled system with “homeostasis as the dominant driver and memory modulation as the auxiliary.”

A key insight: These physiological decisions constitute the vast majority of behaviors by volume, but because they are automatic and unconscious, humans rarely regard them as “decisions.” Yet from the perspective of behavioral output, “stopping work to eat because of hunger” is just as much a decision as “choosing Option A after analyzing pros and cons” — the only difference is their source of weights: the former comes from physiological parameters, the latter from memory tags.

5.2 Middle Layer: The Stress Interrupt System — A Veto-Level Hardware Interrupt

When the amygdala detects a life threat exceeding its threshold, the brain completes a global state switch within milliseconds: the prefrontal cortex is forced offline, the deliberative behavioral decision system is suspended, and even the physiological homeostatic system is temporarily overridden (a person in extreme fear does not feel hunger or fatigue). Control is unconditionally transferred to pre-programmed stress response templates — fight, flight, or freeze.

This process is an operating-system-level kernel interrupt: regardless of what the user program (memory-decision system) or background daemon (physiological homeostasis system) is currently executing, once the kernel interrupt fires, everything is immediately suspended. The catecholamine storm → neural gain elevation → channel amplification → dense encoding initiation — but the prefrontal cortex’s rational evaluation function is shut down. Behavioral output is entirely animal: no sorting, no weighing, no comparison of options.

5.3 Top Layer: The Deliberative Behavioral Decision System — A Uniquely Human Capability Based on Memory Weight Sorting

This layer can only operate normally when bottom-layer physiological needs are basically met and the middle layer has not detected an acute threat. This is the core decision-making capability that distinguishes humans from other animals — human “thinking” is essentially the process of invoking the long-term memory store, sorting by emotional weight tags, and biasing choice preferences through somatic markers. System 1’s “intuition” is the automated output of this process, and System 2’s “deliberation” is the prefrontal cortex’s review and potential override of System 1’s output. All human decisions that require “thinking it over” — career choices, investment judgments, interpersonal strategies, moral reasoning — run on this layer.

Triple-Layer Decision Architecture (Priority from Highest to Lowest)

Layer 1: Stress Interrupt (Highest Priority)
  Amygdala detects life threat → vetoes all other layers
  → Prefrontal cortex goes offline + homeostasis paused
  → Pure animal reflexive output (fight/flight/freeze)
  Trigger frequency: Extremely low (may occur only a few times in a lifetime)

Layer 2: Physiological Homeostatic Decisions (Second-Highest Priority)
  Hypothalamus/brainstem continuously monitors homeostatic parameters
  → Hunger / thirst / fatigue / temperature / pain / sexual drive
  → Automatic behavioral drive when physiological weight reaches threshold
  → Can be temporarily suppressed by willpower, but cannot be ignored long-term
  Trigger frequency: High (likely the largest share of total behaviors)

Layer 3: Deliberative Behavioral Decisions (Based on Memory Weight Sorting)
  Prerequisite: Physiological needs basically met + no acute threat
  → Invoke long-term memory → weight-based sorting → somatic markers
  → System 1 output (intuition) → System 2 review (deliberation) → final decision
  → Decision outcome feedback written back to memory store
  Trigger frequency: Dozens per day (but quality determines life trajectory)

5.4 Priority Logic Among the Three Layers

The three-layer system does not involve “competition” — it follows a strict priority interrupt hierarchy:

The stress layer vetoes everything. A person who has gone without food or water for days (severe physiological layer alarm), if suddenly encountering a predator (stress layer triggered), will still run with full force — physiological needs are instantly overridden.

The physiological layer continuously constrains the deliberative layer. An extremely hungry person cannot make high-quality investment decisions — because weight signals from the physiological layer continuously interfere with the operation of the deliberative decision system. Maslow’s hierarchy of needs essentially describes this hierarchical constraint: when physiological needs are unmet, the upper-layer deliberative behavioral decision system cannot obtain sufficient computational resources.

The deliberative layer runs only within “remaining bandwidth.” This is why the quality of human deliberative decisions depends critically on physiological state — adequate sleep, nutrition, and sense of safety are not “nice to have” luxuries but necessary preconditions for normal operation of the deliberative behavioral decision system. Returning to our earlier discussion of recovery capacity: high-quality sleep, high-quality nutrition, adequate oxygen, and deep meditation are all essentially ensuring that the bottom-layer physiological system is fully satisfied, thereby releasing maximum “cognitive bandwidth” for the upper-layer deliberative behavioral decision system.

Two Boundary Cases: vmPFC Lesions and HSAM

This framework can be examined through two clinical boundary cases — but note that their evidence strengths differ. vmPFC lesions constitute a relatively strong boundary case, while HSAM is a weaker, suggestive case.

The shared result of both is decision paralysis — but for precisely opposite reasons. HSAM is “all weights equal = no weights,” while vmPFC damage is “weights exist but cannot be read.” Both cause the sorting system to fail to produce meaningful output. This “different causes → same consequence” structure provides powerful cross-validation for this framework.

Decision Feedback Loop: The Self-Reinforcing Spiral of Weights

The mapping from memory weights → decision weights is not unidirectional but a closed loop. Decision outcomes are written back to the memory store through new emotional tags, updating weight distributions and altering the output of future analogous decisions — forming a self-reinforcing spiral.

7.1 Positive Spiral: The Formation of Expertise

Successful decisions in a given domain → positive emotional tags → high-weight memory inscription → preferential retrieval in future analogous decisions → greater inclination toward similar choices → more successful experiences → stronger positive weights → the gradual formation of “intuition” and “expertise” in that domain. This is why experienced chess players “see” the best move at a glance — their memory weight tables have undergone tens of thousands of write-and-optimize cycles in that domain.

7.2 Negative Spiral: Trauma and PTSD

Traumatic experience → extremely high negative emotional tag → ultra-high-weight memory inscription → highest-priority retrieval in any related context → avoidance behavior / hypervigilance → deprivation of opportunities for new neutral or positive experiences → the weight table becomes dominated by traumatic memories → gradual generalization to an ever-widening range of contexts → PTSD.

The essence of PTSD is not “thinking too much” — it is that the priority of the traumatic memory has been set abnormally high within the weight system, causing it to be preferentially retrieved in every related decision, distorting the entire decision output. The decision system itself is not broken — a single ultra-high-weight record in the reference database has contaminated the results of all related queries.

7.3 Active Decision-Making Itself Strengthens Memory

Research by Murty et al. (2015, 2019) provides direct evidence that decisions strengthen memory: information revealed through participants’ active choices was remembered better than information pre-assigned by experimenters, and this memory enhancement persisted after a 24-hour delay — a behavioral marker that the consolidation process has been upregulated. Neuroimaging data showed enhanced striatal activation during active choice, and the strength of choice-related striatal activation predicted subsequent increases in hippocampal–perirhinal cortex (PRC) functional connectivity. [Evidence level: moderate-strong, human fMRI + behavioral experiments]

This means: you don’t merely “use memory to make decisions” — you also “use decisions to strengthen memory.” Every active choice reshapes your weight table by triggering the striatal–hippocampal consolidation pathway. This constitutes a self-reinforcing closed loop: high-weight memories drive decisions → decision outcomes are written back to the memory store through new emotional tags → the weights of related memories are further strengthened → the next analogous decision is biased even more strongly in that direction.

This is also where the deeper risk of cognitive offloading to AI lies — when the decision-making process is outsourced, one loses not only the training opportunity of the current decision but also the consolidation pathway through which decisions update and strengthen the memory weight table. Cognitive offloading interrupts not “thinking” itself, but the feedback loop of “decision → striatal activation → consolidation enhancement → weight update.”

Application Insights: Speculative Engineering Extrapolation for AI Alignment

The following discussion presents speculative application insights based on this paper’s neuroscience hypothesis and does not participate in direct validation of the core hypothesis. A vast granularity gap exists between biological metaphor and engineering implementation; the following directions are offered for exploratory reference only.

8.1 The Fundamental Problem with Current RLHF

Standard RLHF aggregates all users’ preference data into a single global reward model — this is equivalent to training all models with the “average human weight table.” The result is adequate for the “average person” but insufficiently precise for any specific individual. A 2025 survey on personalized alignment defines this as the core challenge of the field.

8.2 The Architectural Deficiency of System Prompts

In this framework’s language: system prompts attempt to override “long-term memory” with “working memory.” Pre-trained weights (trillions of parameters) correspond to the human long-term memory weight table — hard-coded through decades of experience; system prompts (a few thousand tokens) correspond to human working memory — temporary instructions with limited capacity, whose influence is constrained by long-term memory weights. A sticky note cannot rewrite decades of life experience.

8.3 The Correct Architectural Direction

This framework suggests a deeper solution path:

Human Memory–Decision System           Possible LLM Analogous Architecture (Speculative)
─────────────────────────              ─────────────────────────────────────

Pre-trained weights (shared human knowledge)  →  Base model pre-training (unchanged)
RLHF (average human preferences)              →  General alignment training (unchanged)
Personal memory weight table                  →  Personal LoRA adapter
  Continuously updated but core-stable            Lightweight parameter delta
Sleep triage                                  →  Periodic offline updates
  Consolidate important, forget outdated           Accumulate interaction data → update weights

8.4 The Core Bottleneck: Unobservability of Implicit Weights

The human memory weight table is implicit — users themselves do not know that “the reason I prefer A over B is that a childhood experience attached a negative emotional tag to B.” RLHF captures users’ explicit preference expressions (“which answer do you prefer”), but the implicit weight system driving that preference is invisible.

Correct personalized alignment requires not asking users “what do you want” but reverse-engineering their implicit weight distribution from long-term behavioral patterns — identifying which interactions triggered “high-arousal states” (extended dwell time, repeated revisiting, elevated emotional word density, immediate action taken), as these are the behavioral-level “catecholamine marker” signals.

The Danger of Cognitive Offloading: When the Weight Table Stops Updating

This framework reveals a harm of cognitive offloading to AI that runs deeper than “declining critical thinking”: the cessation of memory weight table accumulation.

Information gathering = the training process of the perceptual system. Information organization = the weight tag inscription during the encoding stage — when you organize information by hand, your brain is attaching emotional tags to each piece of information; when AI organizes for you, those tags are not inscribed. Information ranking = memory comparison and construction of the decision weight table — when AI ranks for you, your memory store is not invoked and no new weights are inscribed.

A 2025 study of 666 participants found a significant negative correlation between frequent AI tool use and critical thinking ability. But analyzed through this framework, what is being damaged is not merely thinking ability — the entire decision system’s reference database has stopped updating.

No new cognitive effort → no emotional arousal → no catecholamine tagging → no high-weight memories inscribed → the decision system’s reference database frozen in its pre-offloading state → continuous degradation of future decision quality.

Falsifiable Predictions

Conclusion: How Memory Weights Shape Your Default Decision Patterns

By unifying Damasio’s somatic marker hypothesis, Kahneman’s dual-system theory, and Shohamy’s episodic memory–decision framework, this paper argues for the following core proposition:

Human decision weights are not independently generated but are the direct mapping of the memory weight table. The contents of the memory store determine the weight table. The weight table determines the ranking. The ranking determines which option is prioritized for presentation to consciousness. The option prioritized for presentation has a high probability of becoming the final decision.

A person’s “personality,” “preferences,” “values,” and “intuitions” — at the neuroscientific level — can all be traced back to the same thing: the statistical distribution characteristics of high-weight memories in their long-term memory store.

The major contributions of this framework include:

Key References

Damasio, A. (1994). Descartes’ Error: Emotion, Reason, and the Human Brain. Putnam.

Damasio, A. (2018). The Strange Order of Things: Life, Feeling, and the Making of Cultures. Pantheon.

Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux.

Shohamy, D. & Daw, N.D. (2015). Integrating memories to guide decisions. Current Opinion in Behavioral Sciences, 5, 85-90.

Rouhani, N. et al. (2023). Multiple routes to enhanced memory for emotionally relevant events. Trends in Cognitive Sciences.

Dunsmoor, J.E. et al. (2022). Tag and Capture: How Salient Experiences Target and Rescue Nearby Events in Memory. Trends in Cognitive Sciences.

Mather, M. et al. (2016). The Locus Coeruleus-Norepinephrine System: Modulation of Behavioral State and State-Dependent Cognitive Processes. Brain Research, 1709, 1-28.

Clewett, D. et al. (2019). Echoes of Emotions Past: How Neuromodulators Determine What We Recollect. eNeuro, 5(6).

McGaugh, J.L. (2004). The amygdala modulates the consolidation of memories of emotionally arousing experiences. Annual Review of Neuroscience, 27, 1-28.

LePort, A.K. et al. (2012). Behavioral and neuroanatomical investigation of Highly Superior Autobiographical Memory (HSAM). Neurobiology of Learning and Memory, 98(1), 78-92.

Bechara, A. et al. (1997). Deciding advantageously before knowing the advantageous strategy. Science, 275(5304), 1293-1295.

Murty, V.P. et al. (2016). The role of experience in the neural basis of value-based decision making. Current Opinion in Behavioral Sciences, 10, 83-89.

Stickgold, R. & Walker, M. (2013). Sleep-dependent memory triage. Nature Neuroscience, 16, 139-145.

Davis, R.L. & Bhatt, D. (2024). Active forgetting and neuropsychiatric diseases. Molecular Psychiatry.

Shenfeld, I. et al. (2025). Language Model Personalization via Reward Factorization. arXiv:2503.06358.

Gerlich, M. (2025). AI Tools in Society: Impacts on Cognitive Offloading and the Future of Critical Thinking. Societies, 15(1), 6.

Duncan, K.D. & Shohamy, D. (2016). Memory states influence value-based decisions. J. Experimental Psychology: General, 145(11), 1420-1426.

Murty, V.P., DuBrow, S. & Davachi, L. (2015). The simple act of choosing influences declarative memory. J. Neuroscience, 35(16), 6255-6264.

Murty, V.P., DuBrow, S. & Davachi, L. (2019). Decision-making increases episodic memory via postencoding consolidation. J. Cognitive Neuroscience, 31(9), 1308-1317.

Bechara, A. et al. (1997). Deciding advantageously before knowing the advantageous strategy. Science, 275(5304), 1293-1295.

Shields, G.S. et al. (2016). The effects of acute stress on core executive functions. Neuroscience & Biobehavioral Reviews, 68, 651-668.