# Avoidance Learning: Escape, Avoidance, and Why It Persists

> Escape and avoidance learning: signaled and Sidman avoidance, the avoidance paradox, two-factor vs. one-factor theory, and learned helplessness.

- Source: https://operantconditioning.com/avoidance-learning/
- Author: Ryan Martinson (https://operantconditioning.com/about/)
- Publisher: Operant Conditioning Inc.
- Published: 2026-09-09 · Updated: 2026-09-10
- License: https://operantconditioning.com/terms/#copyright (quote with attribution and a link; free to reproduce for non-commercial teaching)

---

*Negative reinforcement · Escape and avoidance*

A dog that learned to jump a barrier after a handful of shocks kept jumping for hundreds of trials without ever being shocked again. How can the absence of something reinforce behavior? The answer took thirty years and explains phobias, procrastination, and safety rituals.

> **Definition**
>
> **Escape learning** is operant learning in which a behavior *terminates* an aversive stimulus that is already present. **Avoidance learning** is operant learning in which a behavior *prevents or postpones* an aversive stimulus that has not yet occurred. Both are forms of [negative reinforcement](https://operantconditioning.com/negative-reinforcement/): the behavior increases because something is removed or kept away.[1]
>
> Shielding your eyes from the sun is escape; putting on sunglasses before you go outside is avoidance. Almost every avoidance response begins its life as an escape response that got earlier.

**In brief**

- Escape ends an aversive stimulus that is already present; avoidance prevents or postpones one that has not yet occurred, and both are [negative reinforcement](https://operantconditioning.com/negative-reinforcement/).
- In successful avoidance the consequence is that nothing happens, yet avoidance is among the most persistent behavior known: the avoidance paradox.
- Avoidance persists because a successful avoider never tests the contingency, which is why blocking the response works when simple extinction does not.

## Escape: the easy case

Escape is learned quickly because the organism feels the contrast directly: the shock is on, the lever is pressed, the shock is off. Thorndike's cats escaping from puzzle boxes were the first laboratory case, and the relief of a headache after aspirin is the everyday one.[2] Nothing about escape is paradoxical. The consequence is present, immediate, and obviously reinforcing.

## Discriminated (signaled) avoidance

The classic procedure adds a warning signal. A light or tone comes on; a few seconds later a shock begins; a response made during the signal cancels the shock, and a response made after the shock begins ends it. Early trials are escape trials. As learning progresses the response moves earlier, into the signal, and the animal stops receiving shocks altogether: the trials have become avoidance trials.

Richard Solomon and Lyman Wynne ran the definitive version in 1953 with dogs in a two-compartment shuttle box. After a few intense shocks the dogs began jumping the barrier during the signal, their latencies shortened to a second or two, and they continued to jump on trial after trial — hundreds of them — without receiving another shock. Several dogs never received more than a handful of shocks in the whole experiment.[3] When the shock generator was later switched off entirely, the jumping persisted; ordinary extinction barely touched it. What worked best was a glass barrier that physically prevented the jump, so that the dogs had to stay in the compartment and discover that no shock came — most effectively combined with a shock for jumping — and even then several dogs never fully stopped, which led Solomon and Wynne to speak of the "partial irreversibility" of the response.[4][12]

## Free-operant (Sidman) avoidance

In 1953 Murray Sidman removed the warning signal altogether. Rats received brief shocks on a timer — say, every 5 seconds — unless they pressed a lever, in which case the next shock was postponed by a fixed period. Two intervals define the procedure: the **shock–shock (S–S) interval**, the time between shocks if the animal does nothing, and the **response–shock (R–S) interval**, the shock-free period each response buys. Every press restarts the R–S clock.[5][6] With nothing to warn them, rats still learned to press at a steady rate that kept shocks rare, and response rate depended in an orderly way on both intervals.

Sidman avoidance mattered because it seemed to strip the procedure to its bones: no signal, so no fear of a signal, and yet learning. What exactly was being reinforced?

## The avoidance paradox

Reinforcement, by definition, is a consequence that follows a response. In successful avoidance the consequence is that nothing happens. A rat that presses every few seconds never gets shocked, and an event that never occurs cannot follow a response. Worse, the better the animal performs, the less contact it has with the contingency: a perfect avoider could not tell whether the shock generator was still connected. Yet avoidance is among the most persistent behavior known. The theories below are attempts to say what the real consequence is.

## Two-factor theory

O. H. Mowrer's answer, first sketched in 1939 and developed into "two-factor" theory from 1947, is that two learning processes run in sequence.[7][8]

1. **Classical conditioning of fear.** The warning signal is paired with shock and comes to elicit a conditioned emotional response — fear — with the racing heart and freezing that go with it. Neal Miller showed in 1948 that this conditioned fear works as an acquired drive: rats would learn a brand-new response, turning a wheel, simply to get out of a compartment where they had once been shocked.[9]
2. **Operant reinforcement by fear reduction.** The avoidance response terminates the signal, and with it the fear. The response is therefore reinforced — not by the absence of shock, which is nothing, but by escape from an aversive internal state, which is something.

On this account the animal is not "avoiding" in the sense of anticipating the shock; it is escaping the fear the signal produces. Leon Kamin confirmed in 1956 that both parts contribute: rats learned best when a response both turned off the signal and cancelled the shock, and each element alone supported weaker learning.[10] The theory also explains the persistence: because the animal responds early, it never stays with the signal long enough to discover that shock no longer follows, so the fear never extinguishes and neither does the response.

### Problems with two-factor theory

- **Well-trained avoiders show little fear.** Solomon and Wynne's dogs, after the first few trials, looked calm and businesslike. Kamin, Brimer, and Black measured fear of the signal directly, by how much it suppressed food-reinforced lever pressing, and found that fear *declined* as avoidance became well learned — while the avoidance response stayed strong.[11] If fear reduction is the reinforcer, the reinforcer seems to fade while the behavior does not.
- **Sidman avoidance has no signal.** Two-factor theorists replied that the passage of time since the last response, and the animal's own proprioceptive feedback, serve as internal signals — a reasonable move, but one that makes the theory hard to test.
- **Extinction is far too slow.** With the shock off, the signal should lose its fear and the response should fade. In practice avoidance can outlast measurable fear by hundreds of trials.[4][12]

## One-factor theory: the missed shock is the reinforcer

Richard Herrnstein and Philip Hineline argued in 1966 that avoidance needs only one process if "consequence" is understood at the level of rates rather than single events. They built a procedure in which a response did not postpone any particular shock; it merely switched the rat from a schedule delivering shocks at a high rate to one delivering them at a lower rate, and shocks still arrived, unpredictably, after responses. Rats learned to respond anyway. The reinforcer, they concluded, is a **reduction in the overall frequency of aversive stimulation** — something organisms are demonstrably sensitive to.[13][14]

James Dinsmoor pushed the point further: stimuli that reliably accompany the avoidance response — the feel of the lever, the click of a relay, or an explicit "safety signal" — are paired with shock-free time and become conditioned reinforcers in their own right. Add a brief tone after each avoidance response and rats learn faster and respond more.[15] On this view the missed shock is not nothing at all; it is a period of safety, and safety has cues.

## Expectancy and modern hybrid accounts

Martin Seligman and James Johnston proposed in 1973 that avoiders learn two expectancies — "if I respond, no shock; if I don't, shock" — and that the response persists as long as the first expectancy is confirmed, which in successful avoidance it always is.[16] Cognitive language, but the same structural insight as one-factor theory: a well-trained avoider never tests the alternative. Current accounts generally combine Pavlovian fear (which clearly drives acquisition), operant reinforcement by safety and by aversive-rate reduction (which maintains the behavior), and expectancies about the contingency (which explain the immunity to extinction).[17]

## Not every response can be an avoidance response

Rats learn to run or jump to avoid shock in a handful of trials and learn to press a lever to avoid it slowly and unreliably, sometimes never. Robert Bolles explained the discrepancy in 1970 with **species-specific defense reactions**: every species has a small innate repertoire for danger — freezing, fleeing, fighting — and an avoidance response is learned readily only if it is one of these or compatible with them. A rat's natural response to a threatening chamber is to freeze or flee, not to manipulate an object; requiring a lever press pits the contingency against biology.[18] It is the aversive counterpart of the Brelands' "misbehavior of organisms," and a reminder that reinforcement selects from what an animal is built to do.

## Learned helplessness: when escape is never learned

In 1967 Seligman, Steven Maier, and Bruce Overmier gave dogs a series of inescapable shocks in a harness, then placed them in a shuttle box where a jump would end the shock. Dogs that had first received *escapable* shock — or no shock — learned to jump quickly. Most of the dogs that had received inescapable shock did not: they whimpered, lay down, and took the shock, even after occasionally jumping and ending it.[19][20] The decisive comparison was the "triadic design": animals in the two shocked groups received exactly the same shocks, one group's controllable and the other's not. Only uncontrollability produced the deficit. Seligman called it **learned helplessness** and proposed that the animals had learned that responding and outcomes were independent.

Fifty years on, Maier and Seligman reversed the interpretation on the strength of the neuroscience. Passivity in the face of prolonged aversive stimulation turned out to be the *default*, driven by serotonergic neurons in the dorsal raphe nucleus; what animals with escapable shock actually learn is that they have control, and that learning, mediated by the ventromedial prefrontal cortex, inhibits the default and allows escape and avoidance to be learned later.[21] Helplessness is not learned. Control is.

## Why avoidance matters outside the laboratory

Avoidance is the operant engine inside most anxiety problems. Fear may be acquired classically — a panic attack in a supermarket, a dog bite, a humiliating presentation — but it is *maintained* operantly, because avoiding the supermarket, the dog, or the meeting is negatively reinforced by relief, and the avoidance prevents the fear from ever being tested. Paul Salkovskis showed that even subtle "safety behaviors" — gripping the trolley, sitting near the exit, rehearsing every sentence — work the same way: they feel protective, and they keep the catastrophe uncontradicted.[22] Compulsions in obsessive–compulsive disorder are avoidance responses to an internal threat, and the treatment that works, exposure with response prevention, is Solomon and Wynne's cure applied to people: block the response so that the feared outcome can fail to arrive.[23][4]

| Everyday avoidance | Aversive event kept away | Why it persists |
| --- | --- | --- |
| Procrastinating on a hard task | The discomfort of starting | Every postponement is reinforced immediately; the task's real cost arrives much later |
| Checking the stove three times | Imagined fire | The house never burns down, which "confirms" that checking works |
| Never speaking in meetings | Possible embarrassment | Silence is safe every single time; the belief is never tested |
| Leaving early to beat traffic | The jam | The jam is never experienced, so the response never meets extinction |
| Ordering tests a patient does not need ("defensive medicine") | A malpractice suit | The suit never comes, which looks like proof the tests work; the cost lands on someone else |
| Buying insurance | Financial loss | Rational avoidance: the rate of catastrophe really is reduced |
| Streak-keeping in an app | Losing the count | A designed avoidance contingency; the aversive event is manufactured |

Not all avoidance is pathological — sunglasses, seat belts, and vaccines are avoidance, and adaptive. The problem cases are the ones in which the feared event would not occur, or would be tolerable, and the avoidance itself costs more than the thing avoided. The diagnostic question is always the same: has the contingency been tested lately?

> **Three confusions worth clearing up**
>
> **Avoidance is not punishment.** In avoidance a behavior *increases* because it prevents something aversive; in punishment a behavior *decreases* because it produces something aversive. **Escape is not avoidance.** Escape ends a stimulus that is present; avoidance prevents one that is not. **Negative does not mean bad.** Both are negative reinforcement because a stimulus is subtracted, and both strengthen behavior.

## Key takeaways

- Escape terminates an aversive stimulus that is present; avoidance prevents or postpones one that has not yet occurred. Both are negative reinforcement, and almost every avoidance response begins as an escape response that got earlier.
- In successful avoidance nothing happens, so the reinforcer is not obvious, and a perfect avoider never contacts the change when the shock is switched off. That is why avoidance can outlast measurable fear by hundreds of trials and why response prevention works when ordinary extinction does not.
- Two-factor theory says the warning signal is classically conditioned to elicit fear and the response is reinforced by escape from that fear, but well-trained avoiders show little fear and Sidman avoidance has no signal. One-factor theory names the reinforcer as a reduction in the overall rate of aversive events plus the safety signals that accompany responding; modern accounts combine both with expectancies.
- Avoidance is the operant engine inside most anxiety problems: fear is acquired classically but maintained by relief, and safety behaviors keep the catastrophe uncontradicted. Exposure with response prevention blocks the response so the feared outcome can fail to arrive.
- Animals given inescapable shock later fail to escape when they can, and only uncontrollability produces the deficit. Maier and Seligman later reversed the interpretation: passivity is the default, and what is learned is control.

### Check yourself

**Every time a warning tone sounds, a rat jumps a barrier and the shock never comes. Since shock is aversive, is the jumping being punished?**

No. Punishment decreases a behavior by producing something aversive; here jumping increases because it prevents something aversive, which is negative reinforcement in the form of avoidance. Negative does not mean bad: a stimulus is kept away, and the behavior strengthens.

**Solomon and Wynne switched off the shock generator, yet the dogs kept jumping for hundreds of trials. Did extinction fail because the dogs were still terrified?**

Not mainly. Well-trained avoiders show little fear; the response persists because a dog that jumps early on every trial receives exactly what it always received, no shock, so nothing signals that the contingency has changed. What worked was a barrier that prevented the jump, so the dogs had to stay put and discover that no shock came.

**You take a painkiller once a headache has started. Your roommate takes one before a long day at the screen. Which is escape and which is avoidance?**

Taking the painkiller once the headache is present is escape, because the behavior terminates an aversive stimulus that is already there. Taking it beforehand is avoidance, because the behavior prevents an aversive stimulus that has not yet occurred. Both are negative reinforcement.

**Two groups of dogs receive exactly the same shocks in a harness, but only one group can turn them off. Which group later fails to escape in the shuttle box, and what does the modern account say the other group learned?**

The group that could not control the shock: only uncontrollability produces the deficit, which is what the triadic design showed. On the modern account, passivity is the default response to prolonged aversive stimulation, and the group with escapable shock learned that it had control, which inhibits that default and allows escape and avoidance to be learned later.

**Explain it to a friend.** Explain why a safety ritual that "works every time" is the hardest kind of habit to drop, using one of your own habits as the example.

## Frequently asked questions

**What is the difference between escape and avoidance learning?**

In escape learning the aversive stimulus is already present and the behavior terminates it (turning off a shock, taking a painkiller). In avoidance learning the behavior occurs before the aversive stimulus and prevents or postpones it (jumping during the warning signal, leaving early to miss traffic). Both are negative reinforcement.

**What is the avoidance paradox?**

Reinforcement is supposed to be a consequence that follows a response, but in successful avoidance the "consequence" is that the aversive event does not happen. Something that never occurs cannot follow anything. Theories of avoidance are attempts to identify the actual reinforcer: escape from conditioned fear (two-factor theory), reduction in the overall rate of aversive events and the safety signals that accompany responding (one-factor theory), or confirmed expectancies.

**What is two-factor theory?**

Mowrer's proposal that avoidance involves two learning processes: first the warning signal is classically conditioned to elicit fear, then the avoidance response is operantly reinforced because it terminates the signal and the fear. It explains acquisition well and persistence partly, but well-trained animals show little fear, and avoidance can be learned without any signal.

**What is Sidman avoidance?**

Free-operant avoidance, introduced by Murray Sidman in 1953: shocks arrive on a timer unless the animal responds, and each response postpones the next shock for a set interval. There is no warning signal. Rats learn to respond at a steady rate that keeps shocks rare, which was hard for signal-based theories to explain.

**Why is avoidance so hard to extinguish?**

Because a successful avoider never experiences the change. If the shock generator is switched off, an animal that responds early on every trial receives exactly what it always received — no shock — so nothing signals that responding is now unnecessary. Extinction requires contact with the new contingency, which is why response prevention (blocking the response) works when simple extinction does not.

**Is learned helplessness still an accepted theory?**

The phenomenon is solid — animals and people exposed to uncontrollable aversive events later fail to escape controllable ones — but the explanation has been revised by its own authors. Maier and Seligman (2016) concluded that passivity is the brain's default response to prolonged aversive stimulation, and that what is learned is control, which inhibits the default. The practical lesson is unchanged: experiences of control are protective.

## References

1. Skinner, B. F. (1953). *Science and Human Behavior*. Macmillan.
2. Thorndike, E. L. (1898). Animal intelligence: An experimental study of the associative processes in animals. *Psychological Review Monograph Supplement, 2*(4), 1–109. Read the 1898 monograph as Chapter II of the 1911 book in the library: https://operantconditioning.com/library/thorndike-animal-intelligence/#chapter-ii-animal-intelligence-an-experimental-study-of-the-associative-processes-in-animals
3. Solomon, R. L., & Wynne, L. C. (1953). Traumatic avoidance learning: Acquisition in normal dogs. *Psychological Monographs, 67*(4), 1–19.
4. Solomon, R. L., Kamin, L. J., & Wynne, L. C. (1953). Traumatic avoidance learning: The outcomes of several extinction procedures with dogs. *Journal of Abnormal and Social Psychology, 48*(2), 291–302.
5. Sidman, M. (1953). Avoidance conditioning with brief shock and no exteroceptive warning signal. *Science, 118*(3058), 157–158.
6. Sidman, M. (1953). Two temporal parameters of the maintenance of avoidance behavior by the white rat. *Journal of Comparative and Physiological Psychology, 46*(4), 253–261.
7. Mowrer, O. H. (1947). On the dual nature of learning — a re-interpretation of "conditioning" and "problem-solving." *Harvard Educational Review, 17*, 102–148.
8. Mowrer, O. H. (1960). *Learning Theory and Behavior*. Wiley.
9. Miller, N. E. (1948). Studies of fear as an acquirable drive: I. Fear as motivation and fear-reduction as reinforcement in the learning of new responses. *Journal of Experimental Psychology, 38*(1), 89–101.
10. Kamin, L. J. (1956). The effects of termination of the CS and avoidance of the US on avoidance learning. *Journal of Comparative and Physiological Psychology, 49*(4), 420–424.
11. Kamin, L. J., Brimer, C. J., & Black, A. H. (1963). Conditioned suppression as a monitor of fear of the CS in the course of avoidance training. *Journal of Comparative and Physiological Psychology, 56*(3), 497–501.
12. Solomon, R. L., & Wynne, L. C. (1954). Traumatic avoidance learning: The principles of anxiety conservation and partial irreversibility. *Psychological Review, 61*(5), 353–385.
13. Herrnstein, R. J., & Hineline, P. N. (1966). Negative reinforcement as shock-frequency reduction. *Journal of the Experimental Analysis of Behavior, 9*(4), 421–430.
14. Herrnstein, R. J. (1969). Method and theory in the study of avoidance. *Psychological Review, 76*(1), 49–69.
15. Dinsmoor, J. A. (2001). Stimuli inevitably generated by behavior that avoids electric shock are inherently reinforcing. *Journal of the Experimental Analysis of Behavior, 75*(3), 311–333.
16. Seligman, M. E. P., & Johnston, J. C. (1973). A cognitive theory of avoidance learning. In F. J. McGuigan & D. B. Lumsden (Eds.), *Contemporary Approaches to Conditioning and Learning* (pp. 69–110). Winston-Wiley.
17. Krypotos, A.-M., Effting, M., Kindt, M., & Beckers, T. (2015). Avoidance learning: A review on theoretical and experimental approaches. *Frontiers in Behavioral Neuroscience, 9*, 189.
18. Bolles, R. C. (1970). Species-specific defense reactions and avoidance learning. *Psychological Review, 77*(1), 32–48.
19. Seligman, M. E. P., & Maier, S. F. (1967). Failure to escape traumatic shock. *Journal of Experimental Psychology, 74*(1), 1–9.
20. Overmier, J. B., & Seligman, M. E. P. (1967). Effects of inescapable shock upon subsequent escape and avoidance responding. *Journal of Comparative and Physiological Psychology, 63*(1), 28–33.
21. Maier, S. F., & Seligman, M. E. P. (2016). Learned helplessness at fifty: Insights from neuroscience. *Psychological Review, 123*(4), 349–367.
22. Salkovskis, P. M. (1991). The importance of behaviour in the maintenance of anxiety and panic: A cognitive account. *Behavioural Psychotherapy, 19*(1), 6–19.
23. Meyer, V. (1966). Modification of expectations in cases with obsessional rituals. *Behaviour Research and Therapy, 4*(4), 273–280.


## About the author

Ryan holds a master's degree from UCLA, where he studied animal behavior in Daniel Blumstein's lab and was part of the university's Evolutionary Medicine Program, which applies findings from evolutionary biology and animal behavior to human health. He founded Operant Conditioning Inc. and built the Operant habit app (https://operantconditioning.com/app/). Every page here is written from the primary literature and cites it. How pages are checked: https://operantconditioning.com/about/#editorial-standards

## Related

- [Negative reinforcement](https://operantconditioning.com/negative-reinforcement/): The quadrant avoidance belongs to, and why it is not punishment.
- [Operant vs. classical](https://operantconditioning.com/operant-vs-classical-conditioning/): Two-factor theory is where the two kinds of learning meet.
- [Extinction](https://operantconditioning.com/extinction/): Why avoidance resists it, and what response prevention does.