Learned Helplessness and Perceived Control

When actions repeatedly appear unable to influence outcomes, people may learn more than disappointment. They may begin to expect that action itself will not matter.

Someone repeatedly uses the actions that appear available, yet the relevant outcome remains controlled elsewhere. Requests receive no response. Following the stated procedure changes nothing. Additional effort produces the same result.

Later, the environment may contain a new route to influence. Earlier experience may still shape whether that route is noticed, trusted or attempted. Whether it is visible, accessible, safe and genuinely effective remains a separate question.

This does not mean the person was wrong to stop acting before the circumstances changed. If their actions genuinely had no effect, reduced effort may have been accurate and protective. Continued non-action may also remain appropriate if the supposedly new route is unsafe, obscure or ineffective.

Behaviour alone cannot establish that an earlier control expectation is driving the response.

The learned-helplessness Model addresses a narrower possibility: experience with uncontrollability can alter later action and perceived control, sometimes allowing earlier conditions to continue influencing behaviour after some control becomes available.

Learned helplessness is the name of a Model, not a description of a person's essence. It does not mean someone has a helpless personality, no longer cares or lacks the character to act.

The name comes from the original interpretation of historical experimental findings. Contemporary accounts retain the importance of controllability but question whether helplessness itself is what is learned.

Uncontrollability is a relationship

Uncontrollability is not simply an unpleasant event or an unsuccessful attempt. It is a relationship between actions and outcomes.

The central question is:

Does changing an available action reliably change what happens?

If altering an action changes the probability, timing, intensity or form of an outcome, the person has some control relative to that action. They may be able to reduce an event's severity without preventing it, delay it without eliminating it or influence one stage while leaving the wider system untouched.

If the outcome remains the same regardless of the actions the person can access, the situation is uncontrollable in that respect.

This distinction connects the Model to the action–outcome learning architecture established in D5.4.

Failure alone does not demonstrate uncontrollability. One action may fail because it was poorly selected, insufficiently practised, interrupted too early or directed toward the wrong part of the problem. Another accessible action may still work.

Adversity is also different. An outcome can be painful or unwanted while remaining partly controllable. Conversely, a neutral event can be impossible to influence.

Predictability is another separate property. Someone may predict an outcome accurately without being able to change it. An outcome may also be difficult to predict while still responding to action.

A genuine contingency may be hard to detect when its effects are small, delayed or probabilistic. The absence of an obvious result does not always prove the absence of control.

The phrase available action matters. A response that could theoretically affect an outcome does not provide meaningful control if the person cannot perceive, access or perform it. Control is relative not only to the environment, but to the person's actual options and capacities.

The Model has eight linked components

The learned-helplessness Model can be organised into eight components:

• outcomes are experienced as uncontrollable;
• the system registers or infers weak action–outcome contingency;
• expectations of control may change;
• initiation may decline;
• persistence may decline;
• exploration may decline;
• the expectation may generalise beyond the original context;
• sufficiently credible action–outcome evidence may support later updating.

Not every case follows every component. The behavioural changes may occur separately, and several mechanisms may contribute to the same observed response.

Learning also need not take the form of a conscious verbal belief such as Nothing I do matters. It may appear through action selection, control judgments, physiological regulation or changes in how effort is allocated.

Histories of control change later responding

The learned-helplessness Model emerged from animal experiments conducted in the 1960s.

In the original studies, animals were exposed to severe aversive events. Some could terminate the event through an action. Others received equivalent stimulation but could not affect when it ended. A substantial proportion of animals previously exposed to uncontrollable stimulation later showed disrupted escape responding in another setting where escape was possible.

These procedures caused significant animal suffering and belong to a historical ethical context. Their findings cannot be transferred directly to complex human life.

The original interpretation was that exposure to uncontrollability taught the animal that its responses did not matter. That learned expectation was then thought to interfere with later escape.

Later neuroscience materially revised this account.

Within this research programme, passivity during prolonged aversive stimulation came to be understood partly as a default stress response rather than a response that must itself be learned. Learning that control exists became the central acquired process. Detecting a reliable relationship between action and outcome could recruit regulatory systems that altered later responding. Prior experience of control could also protect behaviour in subsequent adverse situations.

The contemporary revision does not make control irrelevant. It changes what is thought to be learned.

This revised account is grounded largely in animal neuroscience. It does not remove the need to study human expectations, interpretations and social context. Human behaviour cannot be explained by transferring the original animal mechanism—or its later neural revision—unchanged.

Human laboratory studies also measure different outcomes, including persistence, task performance, affect and control judgments. These should not be collapsed into one unified state called helplessness.

The broader conclusion is therefore restrained: histories of controllability can change later responding, but the underlying mechanisms may include stress responses, learned control, action selection, explicit expectations and their interaction.

The connection to D3.8 is relevant here. State and stress conditions can influence which actions remain accessible, how feedback is processed and whether one situation is treated as similar to another.

Low expected control can change action

As established in D8.17 — Action Can Change Motivation, action produces information that can alter later motivation.

When outcomes repeatedly appear unrelated to available action, further responding may become less likely. This may occur through explicit expectations, learned action selection, stress-related regulation or some combination.

A person may still care deeply about the outcome. They may understand what they want and possess some ability to act. What changes is the expected usefulness of available action.

Wanting an outcome is not the same as expecting to influence it. Capacity to perform an action is not the same as expecting that action to work. Opportunity to act is not the same as believing that the opportunity is genuine.

Control expectation also connects with D9.2 — Sense of Agency. When actions reliably produce outcomes, people may experience themselves as causally effective. When outcomes appear independent of their responses, that experience may weaken.

Several forms of reduced action can follow, but they must remain separate.

Reduced initiation

An action may not begin because its expected influence is low.

That does not prove laziness, lack of desire, misunderstanding or disorder. It may reflect prior learning, current stress, fear, exhaustion, inaccessible options or an accurate assessment that the action will not matter.

Reduced persistence

A person may begin but stop sooner when effort does not produce detectable change.

Stopping may reflect low expected control. It may also be rational disengagement from an ineffective, dangerous or excessively costly strategy.

Reduced exploration

The person may test fewer alternatives because the broader environment appears unresponsive.

Where effective alternatives exist, testing fewer of them can reduce the chance of encountering evidence that a different action works, that the environment has changed or that control is partial rather than absent.

These three outcomes need not occur together. Someone may initiate repeatedly but stop early. Another person may persist with one familiar action while exploring few alternatives. A third may explore extensively without expecting any option to succeed.

"Passivity" is therefore too broad to describe the Model's behavioural predictions. The more precise question is which aspect of action has changed, under which conditions and through which combination of processes.

Reduced exploration can preserve an old expectation

The Model contains a possible information-maintenance process.

A history of uncontrollability may reduce expected control. Lower expected control may make exploration appear less worthwhile. If fewer alternatives are tested, less evidence of newly available control may be encountered. An earlier expectation that receives little clear contradictory evidence may remain one influence on behaviour.

Direct human evidence for this complete maintenance sequence remains limited. It should not be treated as an inevitable loop.

Nor is it self-sabotage.

The environment produced the original evidence. The changed contingency may be difficult to detect. Effective actions may remain hidden, socially costly or physically inaccessible. New control may be weak, unreliable or dependent on support that is not consistently available.

Reduced exploration may also be protective. Testing another action can carry financial, emotional, physical or social costs. A person does not become responsible for proving that an environment has changed simply because someone else believes that control is now possible.

This possible maintenance process is informational, not moral. It applies only where accessible control and credible contradictory evidence actually exist.

Learning may generalise

Control expectations formed in one setting may influence behaviour in another, but generalisation is not automatic.

Transfer may become more likely when situations share:

• visible cues;
• social roles;
• emotional meaning;
• authority structures;
• task demands;
• patterns of reward and punishment.

Someone who repeatedly lacked influence in one institution may approach a similar institution with lower expected control. A later situation that shares important cues or evokes a similar state may be treated as more similar to the earlier context.

Yet a learned expectation can remain highly specific. Low perceived control at work does not establish low control in relationships, health or creative activity. Reduced persistence in one kind of task does not prove global helplessness.

Human reformulations of learned helplessness tried to explain this variation through causal interpretation. They asked whether the source of non-control was understood as:

• internal or external;
• stable or temporary;
• global or specific.

These distinctions became influential, particularly in research on depression. Their evidence is mixed, and their causal role remains contested. They do not provide a complete account of when helplessness generalises.

The safer conclusion is narrower:

Expectations formed in one setting may affect another when the contexts are experienced as sharing relevant causes or cues.

Generality must be demonstrated rather than assumed.

Perceived and objective control can diverge

The Model depends partly on perceived control, but perceived control is not necessarily accurate.

A person can underestimate control. They may continue to expect that action will not matter after a reliable response has become available.

They can also estimate low control accurately. An institution, illness, dangerous relationship or economic condition may genuinely leave few effective options.

Perceived control can be too high as well. Someone may attribute a favourable outcome to their action even when the outcome would have occurred anyway. Frequent action and frequent reward can coincide without a reliable causal relationship.

This matters because perceived control can influence motivation independently of its accuracy. Feeling influential may increase willingness and persistence. But persistence based on illusory control can waste resources, prolong exposure to harm or prevent disengagement from an ineffective strategy.

Higher perceived control is therefore not inherently better.

Control may also be narrow. A person may accurately influence one component while lacking control over wider causes or final outcomes. Partial control should not be described as total agency, but neither should it be dismissed as no control.

Any interpretation therefore requires two separate questions:

• What control does the person perceive?
• What control is actually available relative to accessible actions?

Neither answer can be inferred from action or non-action alone.

Potentially corrective evidence must be credible

If control expectations can change, later action–outcome experience may support further updating. But one successful outcome does not necessarily provide sufficient evidence.

Potentially corrective evidence must show a sufficiently clear and reliable relationship between an accessible action and a relevant outcome.

Updating may remain limited when:

• the effect is too small to detect;
• feedback arrives too late;
• success could be attributed to chance;
• another person appears responsible;
• control works only once;
• the outcome remains mostly determined elsewhere;
• the new context feels unrelated to the earlier one;
• the action cannot be repeated safely;
• capacity remains constrained.

Even clear evidence does not guarantee rapid or complete updating.

Historical animal work reported that prior control could protect later responding and that repeated guided escape responses altered behaviour after uncontrollable exposure. The historical procedure has no direct prescriptive translation to human support or treatment.

The Model does not instruct people to expose themselves to adversity, persist through repeated failure or test known constraints. Action cannot create control where no contingency exists. A person does not need to repeatedly demonstrate that danger, pain or structural restriction is real.

Potentially corrective evidence concerns what an environment reliably allows, not how strongly someone believes they should keep trying.

The Model is not a verdict about the person

Learned helplessness is not a diagnosis.

It is not a complete explanation of depression, trauma, executive dysfunction or any other mental-health condition. Human experimental effects are heterogeneous, and not every exposure to uncontrollability produces later behavioural impairment.

Some controlled studies find changes in affect, perceived control or responding after uncontrollable stress. Others do not find the predicted transfer into later learning. This variability is part of the evidence.

Reduced action can have many causes:

• physical incapacity;
• pain;
• fatigue;
• threat;
• depression;
• inaccessible design;
• insufficient resources;
• rational cost assessment;
• lack of effective options;
• competing values;
• social or institutional control.

The Model is most relevant when a history of controllability plausibly contributes to the pattern, while recognising that multiple processes may operate together.

It must never be used to blame people for adaptations shaped by harmful environments.

Expectations of low control may originate in situations where control truly was absent. Those expectations may remain accurate. Even when circumstances change, the burden of making new control clear and accessible does not rest solely on the person who previously lacked it.

Neither persistence nor disengagement is inherently healthy. The relevant question is whether accessible action can meaningfully influence the outcome under present conditions.

Model boundary and continuation

The governing proposition of this Model is conditional:

Learning that outcomes are uncontrollable can reduce action even when control later becomes possible.

The Model does not claim that this happens universally or through one settled mechanism. It identifies a possible architecture connecting experienced uncontrollability, contingency learning, control expectations, reduced action, generalisation and later updating.

Its strongest contribution is not the idea that some people become permanently helpless. It is the recognition that action depends partly on learned and experienced relationships between behaviour and outcome.

The theory's history, reformulations, animal-to-human translation and current evidential standing require a separate assessment.

That assessment continues in D9.14 — Learned Helplessness: Historical Theory and Current Evidence.