Emotions and impulsivity in a trader's decisions
After a losing trade, you can feel frustrated and still execute the next order according to your plan. After a large gain, you might instead increase the position or skip a required entry condition. What reveals execution quality is the change in the decision, rather than the name of the emotion. This article separates emotion, impulse and action, identifies observable signs of a departure from the plan and shows how to check them in the order history. The aim is to limit the influence of a temporary state on the execution of established rules, without promising to eliminate emotions or improve financial results.

Check the action before explaining it with an emotion
A note saying “I felt tense” and an order sent too early are different observations. The first reports a state; the second can be compared with a prior rule. Either may be worth recording, but a feeling alone does not establish that an entry was impulsive. Calmness alone does not establish compliance either.
Consider an invented replay protocol. A previous position closes at 09:10 UTC. New entries may be sent no earlier than 6 minutes afterward, only after a completed checklist is recorded and only at a volume of at most 0.15 lot. The earliest permitted time is therefore 09:16 UTC. All three conditions must hold. These are arbitrary exercise conditions, not an optimal pause or suggested position size. The self-reported tension scale runs from 0 to 10; it is a private diary field, not a clinical measurement or an order-permission rule.
Classify the records without turning missing evidence into a pass
| Entry | Submitted | Tension note | Checklist recorded before submission | Volume |
|---|---|---|---|---|
| A | 09:14 | 8/10 | Yes | 0.10 lot |
| B | 09:17 | 8/10 | Yes, at 09:16:50 | 0.15 lot |
| C | 09:20 | 2/10 | Record missing | 0.12 lot |
Entry A violates the time condition even though its checklist and volume pass. Entry B satisfies all three stated conditions despite the high tension note. Entry C passes time and volume, but its checklist status is unresolved. Missing evidence is not proof that no checklist was completed, and it is not enough to confirm compliance. Seek the record; until then, keep the case in a separate category.
The same rating in A and B does not mean that their internal states were identical. Nor can the lower note in C repair the missing record. The diary and the order history answer different questions. No trade results are needed to classify these particular conditions, and no results have been supplied.
Turn an intended response into a checkable instruction
“I will be more disciplined” leaves the action unspecified. A candidate instruction is more concrete: if the earliest entry time has not arrived or the checklist record is absent, do not submit the entry and record which condition remains open. This links an observable situation to an action. It does not require first proving what emotion produced the urge.
Gollwitzer and Brandstätter studied implementation intentions as links between situations and intended actions in goal-pursuit tasks. That research is a source for the planning concept, not a test of our pause duration or a demonstration of improved Forex returns. A proposed instruction can also be cumbersome or fail in practice. Retain its version, the opportunities to use it and cases where submission bypassed it; a written sentence is not evidence of an effective control.
Keep observation separate from diagnosis and causation
The replay identifies one confirmed violation, one confirmed compliant entry and one unresolved entry. It cannot show whether tension caused any action, whether a pause changes behaviour or how the procedure affects financial outcomes. The observations were invented rather than sampled. Do not turn those categories into a label for the person.
Lo and Repin’s trader study measured physiological responses during live sessions. Such measurements differ from a diary rating and do not give this worksheet a universal tension cutoff. Even in a personal diary, a rating of 8 is not established as twice the tension of 4. Use the note to describe what was experienced, then examine the observable execution separately. The detailed lesson retains the research context, competing explanations and limitations needed before making stronger claims about emotion and decision-making.
Three exercises with worked answers
The exercise data are hypothetical. Choose an answer, then check the calculation.
Continue to emotion, impulse and observable action
The detailed sections preserve the illustrations and the research discussion. The replay above supplies a small audit of conditions; it does not diagnose a state or validate a behavioural intervention.
1. Emotion, impulse, action – three different things
The sentence "I lost because I let my emotions get the better of me" sounds like an explanation, but it is only a label stuck on after the fact. It cannot be checked, it cannot be refuted, and no procedure follows from it. To be able to do anything with it at all, that sentence has to be broken into three separate things.
An emotion is a state: a change in feeling, attention and readiness to act, triggered by some event. It can be unpleasant, but in itself it is neither an error nor an edge. Frustration after a loss is a normal reaction to an unfavourable outcome. Tension before an entry may signal that the trader is rating the risk higher than usual – and sometimes he is right.
An impulse is an inclination towards a specific action that appears together with the state: the urge to enter immediately, the urge to increase the position, the urge to delete the protective order. An impulse is not yet anything visible on the account. In itself it does not cost a penny.
An action is an order. It is the only element of the three that leaves a trace: timestamp, instrument, size, level of the protective order, moment of modification. Only at this level can anything be measured, and only here can a countermeasure be designed.
On top of that comes a second distinction, the one that disappears most often in texts about trading psychology: state versus trait. A state is momentary and situational; it passes. A trait is a relatively stable disposition, described in research at the level of differences between people. These are two different things, measured by different methods, and they must not be mixed. This article describes only states and the conditions that intensify them. It does not rule on what a trader "is", and it diagnoses nothing; that is what specialists are for, not a Forex course.
The thesis of this article: emotions are part of the decision process and will stay in it permanently. The problem is not feeling, but the predictable change in behaviour that leads to departing from the adopted process – including when the state is pleasant and the outcome positive.
2. The sequence: stimulus, interpretation, state, impulse, action
Between an event on the chart and a sent order there is more than one step, even though subjectively it looks like a single reflex. Breaking that path into stages has a practical point: each stage is a different place to intervene, and the further from the order, the cheaper the intervention and the less it depends on self-control.
Three remarks about this diagram, without which it is easy to read too much into it.
First, the stages have no fixed durations. Numbers published in guides along the lines of "an emotional reaction takes 200 milliseconds and an analytical one 15 seconds" do not come from any study of traders' decisions. This diagram organises concepts; it does not measure time.
Second, interpretation is a separate step, not an automatic consequence of the stimulus. The same price move is sometimes described as "the setup I was waiting for" and sometimes as "an opportunity I am losing right now". The second description produces time pressure that the first does not have, and time pressure is one of the few conditions in a text about emotions that can be named concretely and checked in the data: how much time passed between the stimulus and the order.
Third, an impulse does not have to end in an action. That is not an optimistic caveat but the condition under which this whole article makes sense. If a state always turned into an order, no procedure could possibly work, and section 7 shows the same state in two different executions.
3. Impulsivity defined by behaviour
The word “impulsivity” in the title concerns a rash transition from impulse to action without the required assessment of conditions. The statement “I felt a strong urge” alone cannot assess a particular order. In the history, we first check observable breaches of previously recorded rules. These may accompany an impulse, but can also be deliberate or result from an execution error. A breach alone does not establish its cause. Practical warning signs include:
- an entry without the required premise – a condition that, under the plan, had to occur before opening;
- a position size set outside the rule the trader applies day to day;
- deleting or moving away the protective order after opening the position, without a rule that provided for it;
- sending an order before finishing your own pre-trade procedure;
- an immediate re-entry after closing a position despite a break that was in force;
- a change of instrument or timeframe to one that is not in the plan for that session.
Each item on this list has two features that make it worth writing down at all: it can be read out of the order history without asking the trader how he felt, and it can be counted across a run. That makes it possible to talk about the frequency of violations instead of about a "weak psyche". The criterion is always the same: the divergence between what was written down before the decision and what was executed, not the mere fact that the trade produced a loss.
State and trait – why they are not the same
In the psychology of individual differences, impulsivity has long since stopped being treated as a single property. Whiteside and Lynam[5] applied factor analysis to the existing impulsivity questionnaires and obtained four distinct factors: urgency (a disposition towards rash action under strong emotion), lack of premeditation, lack of perseverance and sensation seeking. Cyders and Smith[6] in a 2008 review split urgency into negative and positive, that is, a disposition towards rash action under strong unpleasant affect and under strong pleasant affect.
This distinction has one concrete consequence for a trader and one strong limitation.
The consequence: rash action under a pleasant state is a separate phenomenon in this literature, not a footnote. The popular picture in which errors come solely from fear and frustration is incomplete; section 4 returns to this using a run of gains as the example.
The limitation: these are constructs measured with self-report questionnaires, in general and student populations, not in traders during a session. What I carry across from here is only the conceptual distinction – not an effect size and not any forecast of results. Nor does anything follow from it that the reader is supposed to "diagnose" in himself: a personality questionnaire score is not a description of a specific trade, and this article proposes no test.
4. Arousal works after a gain too
Arousal is the dimension of a state that is easiest to confuse with its sign. Euphoria and rage are opposite emotions, but in terms of arousal they lie on the same side. Arousal and emotional valence are different dimensions, and both may matter for action. The state alone does not establish whether execution of the plan will change.
The classic description of that change is Easterbrook's hypothesis[8] from 1959: as arousal rises, the range of cues a person uses on a task narrows. In practice this means attention concentrates on whatever is most salient – the price move and the current result – at the expense of peripheral information: the calendar of data releases, correlations between positions, earlier notes from the plan. This is a hypothesis formulated more than sixty years ago, tested many times, with results that depend on the type of task. I treat it as a frame for organising observations, not as a law.
A run of gains is a state, not proof
After several winning trades in a row a state appears that subjectively reads as accuracy, and that fits ordinary arousal just as well. Arithmetic helps here more than persuasion. At a 55 percent win rate, and assuming successive trades are independent, the probability of five wins in a row within a pre-specified block of five is 0.555, that is about 5.03 percent – roughly one such block in twenty. Across several hundred trades such a run is simply to be expected. There is no information in it that the trader has suddenly started seeing the market better.
The caveat to this calculation matters as much as the calculation itself: the independence assumption is convenient, not true. Real trades are sometimes dependent, because they arise in the same market conditions, on correlated instruments and under the same version of the strategy. The full treatment of runs together with its assumptions belongs to the article on trading under uncertainty (10.1); here all that is needed is the order of magnitude.
The behaviour visible in the data in this state, on the other hand, is entirely concrete: a shorter interval between decisions, a position size outside the previous rule, entries on instruments outside the session plan, dropping part of the procedure as "unnecessary, since it is working anyway". All of these points are on the list from section 3, and none of them requires the trader to admit to euphoria.
5. The hot–cold gap and its design consequence
The plan is written on a Sunday evening or before the session, over coffee, with no position open. It will be executed by someone who is looking at a third loss in a row, or at a move that started without him. Formally it is the same person. In terms of what he will consider reasonable, not necessarily.
Loewenstein[2] described this as the hot–cold empathy gap. In a cold state people systematically underestimate the influence a hot state will have on their behaviour: arousal, hunger, pain, time pressure or a strong emotion. The gap works in both directions. In a hot state, people are equally systematic in failing to reproduce what seemed obvious when cold. Loewenstein later developed this on the ground of medical decisions[3], where the effects are well documented: patients' declarations about their own future choices diverge from the choices made in the real situation.
Both preceding sources are theoretical reviews, rather than single experiments. A related finding concerns the illusion of self-restraint. Across four studies, Nordgren, van Harreveld and van der Pligt[4] examined overestimation of impulse control and exposure to temptation. In study four, they followed 55 people who had stopped smoking at least three weeks earlier. Stronger belief in self-control was associated with less reported avoidance of temptation; less avoidance, in turn, was associated with more smoking after four months. The direct association between belief and subsequent smoking was not statistically significant. These were self-reported observational associations, not proof of causation or a measurement of traders’ behaviour.
The design consequence is single and counter-intuitive. Since the author of the plan, writing from a cold moment, is a poor forecaster of his own behaviour in a hot moment, the plan cannot assume that at the critical moment it will be enough to try harder. It has to assume the opposite: that at the critical moment it will be executed by someone less patient, with narrower attention and with a ready-made narrative justifying an exception.
The whole rest of this section of the course lives off that conclusion. Designing procedures, contingency plans and pre-commitment belong to the article on checklists over willpower (10.7), and ready-made interruption and return protocols to the article on reset protocols (10.9). It is only worth noting what that conclusion does not require: it needs no theory about willpower being depleted. Large replications did not confirm the proposed effect under the protocols examined – in the registered multi-lab replication by Hagger and colleagues[11] (23 laboratories, N = 2141) the effect came out at d = 0.04 with a confidence interval covering zero, and in the later multi-lab test by Vohs and colleagues[12] (36 laboratories, N = 3531) at d = 0.06. The debate remains open: Dang et al. (2025) observed an effect with a longer response-inhibition task, in a project that was only partly preregistered[15]. It did not study trading or establish a universal “willpower fuel”. The hot–cold gap remains a separate rationale for designing procedures.
6. From a state to an observable symptom
The table below does one thing: it converts the names of states into events that can be counted in the order history, and for each countermeasure it states what it rests on. The last column matters more here than usual, because most recommendations in this field are workshop practice without dedicated research, and it is better to write that plainly than to pretend each one has a literature behind it.
| State | Typical change in execution | Symptom in the order history | Countermeasure | Evidential status |
|---|---|---|---|---|
| Frustration after a loss | A shorter path to the next decision, lowered requirements for the premise | Interval from the close to the next order shorter than usual; entry without a condition from the plan | The entry condition must exist in a record predating the stimulus; no record blocks the order | Descriptive mechanism; the countermeasure itself is workshop practice |
| Arousal after a run of gains | Position size outside the rule, entries outside the session plan | Position size deviating from the median of the last few dozen entries; instrument outside the list | Position size determined the same way regardless of recent results | Positive urgency as a construct described in the literature[6], non-trader populations |
| Time pressure on a move that started without you | Shortened analysis, entering on price instead of on a condition | Time from the first move to the order shorter than the normal entry preparation time | Entry allowed only when the condition could have been written down before the move | Craft practice; more in 10.10 |
| Fear of entering after a run of losses | Skipping premises that agree with the plan, shortening the distance of the protective order | Premises marked in the plan with no order after them; protective order distance below the usual | A register of skipped premises – kept separately from the trades actually taken | Craft practice; the register only makes sense across a run of several dozen cases |
| Low stimulation, a long absence of premises | Lowering the "that will do" threshold, entries outside the session window | Orders outside the hours specified in the plan; a rise in the number of entries with no change in the market | Defined windows outside which opening orders are not created | Craft practice, with no research on traders |
| Fatigue towards the end of a long session | A shortened procedure, more decisions per unit of time | A rise in order frequency in the last hour; no entries made before the decision | A limit on the number of decisions, not on time in front of the screen | Physical variables have their own article – 10.11; no causal conclusions here |
Three caveats about this table, without which it turns into yet another list of tips.
Each of these symptoms has a simpler explanation than emotion, and that has to be ruled out first. A shorter interval between decisions may follow from the market having just entered a period in which there are simply more premises. A deviating position size may result from different instrument volatility. More orders in the last hour of a session is often just an effect of the hour, not of fatigue. The emotional label may be attached only once the market conditions, the instrument and the time of day have been compared – not when it fits the narrative.
No countermeasure in this table works because the trader has resolved that it will. Each one is a rule with four fields: trigger condition, action, completion criterion, audit trail. That form is defined by the article on checklists over willpower (10.7), and the principles for designing it live there; here it is only applied. Section 8 shows it on two procedures written out to the end, together with a separately added criterion for the rule not working.
And the most important thing: the table does not say that trading is forbidden in these states. It says where to look for the trace when execution drifts away from the plan. A ban on trading "above an agreed stress threshold" would be a rule based on self-report, which this scale cannot carry; section 8 returns to that.
7. The same emotion, two executions
The most common failing of texts about emotions in trading is that every state they describe ends in a loss. The reader comes away convinced that feeling is equivalent to error – which is exactly the opposite of what follows from the data. That is why the same starting state is shown below in two executions, one of them additionally in two outcome variants.
The shared starting point. In one session two trades consistent with the plan end in a loss, each at the risk assumed in the plan: −1.0R and −1.0R. The trader notes clear frustration and an urge to win it back immediately. The condition for a third entry from the plan does not occur at that moment.
Execution A. The trader does not enter, because the condition is not there. The session ends at −1.0R −1.0R = −2.0R. The emotion was strong, and there was not a single violation. This is a correctly executed session, despite the loss and despite it not having been pleasant.
Execution B. The trader enters despite the absence of the condition and doubles the risk relative to the plan. The trade ends in a loss: −2.0R. The session closes at −1.0R −1.0R −2.0R = −4.0R. Two violations from the list in section 3: an entry without a premise, and a position size outside the rule.
Execution B′. The same two violations, a different draw. The trade ends in a gain: +2.0R. The session closes at −1.0R −1.0R +2.0R = 0.0R. The account looks better than in execution A.
Putting these three runs side by side says two things at once.
First: the assessment of the execution cannot depend on which variant came up. In executions B and B′ the behaviour is identical, and the difference in outcome is 4.0R and comes purely from the draw. Variant B′ is at the same time more dangerous pedagogically than B, because the violation was rewarded – and at the next similar moment it will be available as a precedent. The full development of this thread, together with the process-and-outcome matrix, belongs to the article on outcome versus process (10.4).
Second: execution A is a normal, correct course for a session, even though the emotion in it was just as strong as in the others. A strong emotion without a process violation is not a theoretical possibility or an exception granted to a chosen few. It is a possible course for a session that should be distinguished from a breach, rather than treating feeling itself as an error. A text that does not show this teaches the reader that the only way out is to stop feeling.
One thing to straighten out while we are here, because it comes up most often in this context. The sentence "a loss hurts exactly twice as much as an equal gain pleases" is too strong: the asymmetry of losses and gains is documented[13], but it is not a constant and it has serious critics[14], and the value being referred to is a parameter of a function, not a multiplier to be applied linearly. The size of that coefficient, its interval, the criticism and the consequences for decisions about holding positions belong entirely to the article on the psychology of losses (10.6).
8. How to observe this in yourself without fooling yourself
Everything above is useless until a trader has a way of checking whether it is his case at all. And this is where the most common methodological error in this field begins: drawing conclusions from your own dozen or so trades.
Why twenty trades are not enough
The typical observation goes: "I split my trades into calm and nervous ones; in the calm ones I have six hits out of ten, in the nervous ones four, so emotions cost me twenty percentage points of accuracy." Let us check what a difference like that is worth.
Assume a situation maximally unfavourable to that hypothesis: both groups have an identical true accuracy of 50 percent, ten trades each, outcomes independent. The probability that the numbers of hits in the two groups will differ by at least two, that is, that we will see "60 versus 40" or more, is:
- P(difference = 0) = C(20,10) / 220 = 184,756 / 1,048,576 = 0.1762
- P(difference = 1) = 2 × C(20,11) / 220 = 335,920 / 1,048,576 = 0.3204
- P(difference ≥ 2) = 1 − 0.1762 − 0.3204 = 0.5034, that is about 50 percent
Simulation check: 400,000 draws, seed 20260810, result 0.5038, consistent with the calculation. In other words: with two groups of ten trades, a "60 versus 40" difference shows up in roughly half of all cases even when there is no difference at all. That difference alone is insufficient to conclude that the state changed the hit rate. That does not mean the effect is absent; it means these data can neither show it nor rule it out.
Two further problems remain even with a large sample. First: the state is usually assigned after the outcome is known, and memory reconstructs. Second: the state is not the only thing distinguishing the two groups – nervous trades fall more often at different hours, on different instruments and at different volatility. Methods of comparing runs, violation codes and process metrics in a trading journal belong to the article on the trading journal (10.8); here all that is needed is the awareness that without a record made before the decision there is nothing to compare.
The limits of a self-report scale
The "my state, 1–10" scale is useful, but for one thing only: comparing your own entries with each other. The difference between 6 and 7 is not a psychological quantity and does not have the same content for two different people, or for the same person six months later. A missing entry does not mean an absence of emotion, only an absence of an entry. And a rule along the lines of "I do not trade at a state above six" has a flaw built into its construction: the assessment is issued by the very state it is supposed to limit.
The same goes for measurements taken from the body. An elevated heart rate before an entry is data about a state and nothing more. It does not say whether the entry is a good one, and the studies cited here do not establish a link between such a reading and decision quality in retail traders. It may be recorded as one indicator of state. It must not be turned into a decision threshold, nor may you promise yourself that "the correlation will show up after a week".
Two procedures written out to the end
The two rules below belong to this article because they concern only the moment between the state and the order. Everything broader (designing the plan, checklists, technical pre-commitment) belongs to the article on checklists over willpower (10.7) and to the trader checklists.

Procedure 1: the sentence before the order.
Trigger condition: every order that opens a position.
Action: before sending the order, write one sentence in a fixed format: "I am entering because [condition from the plan] occurred; risk [value]; I will exit if [invalidation condition]."
Completion criterion: if the sentence can be written in full, the order goes out; if it cannot, the order is not created. There is no third possibility.
Audit trail: a file with the entries and their timestamps, comparable against the order history.
Criterion for the rule not working (a requirement separate from the form itself): after several dozen entries the share of vague sentences ("looks good", "momentum") does not fall; or the timestamps of the entries are later than the orders, which means they are being produced as justification rather than as a condition.
Procedure 2: the register of entries not taken.
Trigger condition: the occurrence of a premise from the plan that was not followed by an order.
Action: record the premise, the time and one word about your state – before it is known what price did next.
Completion criterion: the entry is closed at the moment the premise ceases to apply.
Audit trail: the number of entries per week and their distribution relative to the outcomes of previous trades.
Criterion for the rule not working (a requirement separate from the form itself): the register is empty in weeks with losses and full in calm ones – that is, it appears when it is easy and disappears when it is needed.
Both procedures share the same flaw, and it is worth naming: they are voluntary. Nothing blocks sending an order without the sentence and nothing forces an entry into the register. Hard blocks that cannot be circumvented in a hot moment are an entirely different class of tool and are the subject of 10.7 and 10.9. Numerical loss limits, position sizes and drawdown thresholds belong neither to this article nor to this section of the course; their owner is section 9 on risk management. Section 10 treats them as external rules whose violation is behaviour worth describing, and does not set their values.
9. What this article deliberately does not do
Some accounts of trader emotions invoke a hormone, a brain region or a reward loop. This article instead relies on behaviour and the measurements reported in the cited papers. Physiological data matter, but they do not identify a specific brain event behind an individual order.
The studies of traders cited in this article did not measure neurotransmitter release during a session. This describes the literature checked as of 8 October 2026; it does not establish that no such studies exist. The claim that “every win releases dopamine into the nucleus accumbens” would require direct evidence that these papers do not provide. Likewise, an amygdala “hijack” is not a mechanism measured here. The metaphor cannot explain why a particular order departed from the plan; that requires a record of the available information and the action taken.
Physiology may accompany decision-making, but the type of measurement limits the conclusion. The following three examples identify the participants, the measurements and the questions their results cannot answer.
These three papers support a cautious conclusion: arousal was observed among professional traders, while its relationship with decisions depends on the task and the variable measured. A state measurement alone does not establish the quality of a particular order. The procedures in this article therefore focus on information, behaviour and execution, without assigning them an unmeasured neurobiological cause.
The model of fast and slow thinking can help organise a description of a decision, but the label “System 1” does not identify what changed in an order. Similarly, “I let my emotions get the better of me” needs a more precise account: which condition was skipped, what information was available and when the action occurred. That account allows you to check whether a proposed procedure addresses the problem identified.
Cognitive biases, the second family of mechanisms that distort decisions, work also when the state is calm and execution matches the plan. They are handled by the article on cognitive biases (10.3). And that is where the usefulness of the sentence "it was done on the spur of the moment" comes to an end.
FAQ – Frequently asked questions
Is it possible to trade without emotions?
Does a strong emotion mean I should not open a position?
How do I tell an impulse from a fast but correct decision?
Is the state after a run of gains more dangerous than after a run of losses?
Is it worth measuring heart rate or recording your state on a 1–10 scale?
How many trades does it take to establish that emotions are costing me money?
Sources and bibliography
- Lo A.W., Repin D.V., "The Psychophysiology of Real-Time Financial Risk Processing", Journal of Cognitive Neuroscience, 14(3), 2002, pp. 323–339. Physiological measurements in 10 institutional traders during real sessions.
- Loewenstein G., "Out of Control: Visceral Influences on Behavior", Organizational Behavior and Human Decision Processes, 65(3), 1996, pp. 272–292. A theoretical review paper introducing the hot–cold empathy gap.
- Loewenstein G., "Hot–Cold Empathy Gaps and Medical Decision Making", Health Psychology, 24(4S), 2005, pp. S49–S56. A review of applications of the gap to medical decisions.
- Nordgren L.F., van Harreveld F., van der Pligt J., "The Restraint Bias: How the Illusion of Self-Restraint Promotes Impulsive Behavior", Psychological Science, 20(12), 2009, pp. 1523–1528. Four studies; the fourth on 55 people after quitting smoking, observed over 4 months.
- Whiteside S.P., Lynam D.R., "The Five Factor Model and Impulsivity: Using a Structural Model of Personality to Understand Impulsivity", Personality and Individual Differences, 30(4), 2001, pp. 669–689. Factor analysis of impulsivity questionnaires; four factors.
- Cyders M.A., Smith G.T., "Emotion-Based Dispositions to Rash Action: Positive and Negative Urgency", Psychological Bulletin, 134(6), 2008, pp. 807–828. The separation of negative and positive urgency.
- Yerkes R.M., Dodson J.D., "The Relation of Strength of Stimulus to Rapidity of Habit-Formation", Journal of Comparative Neurology and Psychology, 18, 1908, pp. 459–482. A study on dancing mice; the source of the later generalisation known as the inverted-U curve.
- Easterbrook J.A., "The Effect of Emotion on Cue Utilization and the Organization of Behavior", Psychological Review, 66(3), 1959, pp. 183–201. The hypothesis that the range of cues used narrows as arousal rises.
- Coates J.M., Herbert J., "Endogenous Steroids and Financial Risk Taking on a London Trading Floor", PNAS, 105(16), 2008, pp. 6167–6172. Observational study, 17 traders, 8 days; no association between cortisol and the level of the result.
- Kandasamy N., Hardy B., Page L., Schaffner M., Graggaber J., Powlson A.S., Fletcher P.C., Gurnell M., Coates J., "Cortisol Shifts Financial Risk Preferences", PNAS, 111(9), 2014, pp. 3608–3613. Randomised, double-blind, placebo-controlled experiment, 36 participants, 8 days of hydrocortisone administration.
- Hagger M.S. et al., "A Multilab Preregistered Replication of the Ego-Depletion Effect", Perspectives on Psychological Science, 11(4), 2016, pp. 546–573. Registered replication report: 23 laboratories, N = 2141, d ≈ 0.04, confidence interval covering zero – an effect indistinguishable from zero.
- Vohs K.D. et al., "A Multisite Preregistered Paradigmatic Test of the Ego-Depletion Effect", Psychological Science, 2021. Multi-laboratory test, 36 laboratories, N = 3531.
- Brown A.L., Imai T., Vieider F.M., Camerer C.F., "Meta-analysis of Empirical Estimates of Loss Aversion", Journal of Economic Literature, 62(2), 2024, pp. 485–516. 607 estimates from 150 articles; mean 1.955, interval [1.82; 2.10].
- Gal D., Rucker D.D., "The Loss of Loss Aversion: Will It Loom Larger Than Its Gain?", Journal of Consumer Psychology, 28(3), 2018, pp. 497–516. A critique of the status of loss aversion as a general law.
- Dang J. et al., “Revisiting Ego Depletion: Evidence from Multi-Lab Collaborations”, Journal of Pacific Rim Psychology, 2025. DOI. 14 samples, N = 2,078; a longer response-inhibition task preceded a Go–Nogo task. Only part of the project was preregistered; the finding does not concern trading.