Predict behavior from consequences and schedules

Predict behavior from consequences and schedules

Classify added and removed consequences by their observed effect, then trace a reinforcement schedule.

A question to investigate

When an unpleasant tone stops after a response and the response becomes more frequent, is that punishment?

Distinguish negative reinforcement from punishment and calculate reward delivery under fixed ratio and interval rules.

Mechanism and evidence

Operant conditioning concerns learning relations between behavior and consequences. Reinforcement increases a behavior; punishment decreases it. Positive consequences add something, while negative consequences remove something. These labels describe function rather than whether a consequence seems kind or unpleasant. Ratio schedules depend on response counts; interval schedules depend on elapsed time followed by a response. Fixed schedules use a constant requirement; variable schedules change that requirement around a pattern or average. These are the introductory distinctions summarized in the linked reading.

The tone is removed and the target response increases, so the invented record illustrates negative reinforcement. Calling it punishment because the tone is unpleasant ignores the observed change in the response. If removing game access after interruptions were followed by fewer interruptions, that different record would illustrate negative punishment. The classification needs a defined behavior and its subsequent frequency.

For R, number the six responses. The third at 7 seconds and sixth at 19 seconds earn rewards. For I, the 2-, 4- and 7-second responses occur before ten seconds. The first eligible response is at 11; the clock then resets. Responses at 12 and 19 occur only one and eight seconds later, so neither earns another reward. A reward does not appear automatically at second 10: the rule requires a response.

Change exactly one feature: add responses at seconds 20 and 22. Predict before annotating. R now has eight responses, so no ninth-response reward yet. I’s next eligible response is at or after second 21; second 20 is early and second 22 earns the next reward. This explicit procedure is a programmed schedule, not a prediction that all learners respond at those times.

A useful transfer explanation distinguishes a schedule from a mechanism of memory or classical conditioning. In classical conditioning a predictive cue becomes associated with another stimulus; the focal response need not earn a consequence. Write a different invented example of each and identify the observation that would separate them. Do not conduct aversive conditioning on people or animals for this activity.

Inspect the relationships

Consequence	Behavior increases	Behavior decreases
Something added	Positive reinforcement	Positive punishment
Something removed	Negative reinforcement	Negative punishment

Timeline: 0 → 2 → 4 → 7 → 11 → 12 → 19 seconds. Mark rewards above the line for R and below it for I.

Worked example

Removing the tone followed by increased responding is negative reinforcement. Rule R rewards at 7 and 19 seconds. Rule I rewards at 11 seconds only in the supplied six-response trace.

Independent application

After a noisy alert is added following an error, errors become less frequent. Which description fits the supplied change? Explain the denominator, process or rule before checking the response. Record a revision after feedback.

Reading

OpenStax Psychology 2e, operant conditioning. The numerical records and tasks are original fictional teaching examples; they are not findings reported by this source.

Compare the rule with the observed behavior

Step through a schedule

Predict the next delivery before stepping. The supplied responses stay fixed across rules; the program does not learn or generate behavior. A delivered token is only a scheduled consequence. Calling it a reinforcer requires evidence that it increases the target behavior.

Consequence rule Fixed ratio: every 3 responsesVariable ratio: requirements 2, 4, 3Fixed interval: first response after 10 secondsVariable interval: waits 5, 12, 7 seconds Add responses at seconds 20 and 22

Response seconds: 2, 4, 7, 11, 12, 19. Variable requirements cycle in the shown order for reproducible teaching; this is not a random schedule. The interval clock starts at zero and restarts on delivery. Equality meets the requirement. Waiting alone never delivers a token.

Next response · 2 s Advance clock 1 second Back 1 second Reset schedule

Clock 0 s · 0 responses processed · 0 tokens delivered.

Current requirement: 3 responses. Since the last delivery: 0 responses.

○ response without delivery · ◆ response with token. Tick positions use the same 0–30 second scale for all rules.

Response at	Requirement	Count since delivery	Consequence

No response has occurred yet.

Changing the rule or response set restarts the clock. Controls are manual; there is no timer. This widget is temporary and resets when the page reloads. Copy your rule, response times and explanation into the source notebook to retain evidence; notebook captures do not save this widget's clock or controls.

Start with fixed ratio and step through the six original responses. Record the delivery times. Change to fixed interval, advance the clock to 10 seconds and explain why reaching the time threshold does not itself deliver a token. Process the response at 11, then add the two late responses and test the reset clock. Changing inputs starts a fresh trace; compare notes across traces.

For variable ratio, the successive count requirements are 2,4,3, then repeat. For variable interval, the successive waits are 5,12,7 seconds, then repeat; each wait begins at the previous delivery. Predict all delivery times before using the controls. These finite repeating lists illustrate changing requirements. They do not reproduce unpredictability or establish the response patterns observed in behavioral research.

Review all four schedule traces

With the original six responses, fixed ratio delivers at 7 and19 seconds; fixed interval at 11; variable ratio at 4 and19; variable interval at 7 and19. In the variable-interval case, the second wait begins at 7 and ends at 19, so the response exactly at 19 qualifies. After it, the next wait is7 seconds. Adding responses at 20 and22 gives fixed interval an additional delivery at 22; it gives none of the other three rules another delivery in this trace.

At clock 21 under extended fixed interval, the 10-second wait since delivery at 11 has elapsed, but no response occurs at 21. Delivery waits until the response at 22. A rule that delivers automatically when time elapses would be a different procedure.

Transfer without the widget

A different fixed-interval rule requires 6 seconds after each delivery. Responses occur at 1,6,7,12 and18 seconds. Write delivery times, then compare with a fixed-ratio rule requiring two responses. Define when the interval clock restarts; do not keep measuring every wait from the original time zero.

Review the unfamiliar timing case

The interval rule delivers at 6,12 and18. The ratio rule delivers at 6 and12; the fifth response at 18 has not met a new two-response count. Same delivery times early in a trace do not prove that two schedules are the same. Use a response pattern that separates their predictions.

A token is not a demonstrated reinforcer merely because a program delivers it. To establish function, define the target behavior and compare its later frequency under an appropriate design. This exercise traces consequences while holding behavior fixed; it cannot show learning, a characteristic response-rate curve, resistance to extinction or a causal effect on people.

Shape a response: change what qualifies

A fictional museum offers an optional map-reading practice game. The target is to point within one centimeter of a marked destination. The learner can stop without losing access to the museum. A coach proposes gradually narrowing the accepted distance instead of requiring perfect accuracy immediately. All observations below are invented; no visitor experiment is being conducted.

Shaping reinforces successively closer approximations to a target behavior. Here, an approximation is a point near the destination. Once the criterion becomes stricter, a formerly accepted point may no longer qualify. This changes which response earns a consequence. A ratio or interval schedule changes when a qualifying response earns one. Both features can be specified together, but they answer different questions.

Try: write the consequence ledger before checking it

The practice protocol delivers one token for every qualifying response. Each row contains four successive attempts, measured in centimeters from the destination. Smaller distances are more accurate. A distance exactly equal to the criterion qualifies. For this exercise, advance to the next criterion only after at least three of four attempts qualify; otherwise repeat the current criterion. This advancement rule is an invented teaching protocol, not a universal recommendation for instruction.

Block	Current maximum distance	Four observed distances, in order	Your token decisions	Your next criterion
A	8 cm	9, 7, 6, 8	Write yes/no for each	Stay at 8 or advance to 4?
B	4 cm	6, 4, 3, 5	Write yes/no for each	Stay at 4 or advance to 1?
C	4 cm	4, 3, 2, 1	Write yes/no for each	Stay at 4 or advance to 1?
D	1 cm	2, 1, 0.5, 1.5	Write yes/no for each	Keep practicing at 1?

Use the source card Shaping criterion and transfer audit to save your ledger. Before opening the review, explain why the 6 cm response in A qualifies but the 6 cm response in B does not. The measured response is identical; the active criterion differs.

Review the shaping ledger

A: no, yes, yes, yes — three tokens, so advance to 4 cm. B: no, yes, yes, no — two tokens, so remain at 4 cm. C: yes, yes, yes, yes — four tokens, so advance to 1 cm. D: no, yes, yes, no — two tokens; the final target has been reached on two attempts but has not become consistent under the stated three-of-four rule. The complete record has 11 tokens across 16 attempts.

The coach should not count A’s 6 cm response as meeting the final 1 cm target. Early approximation success and final-target performance are different outcomes. Nor should the coach advance after B merely because two attempts improved: the declared rule requires three qualifying attempts.

Experiment on the record: change one rule

Re-score the same 16 distances using the final 1 cm criterion from the very beginning. Then re-score them with the initial 8 cm criterion throughout. Compare token totals and explain what neither comparison can establish.

Review the changed criteria

Requiring 1 cm throughout produces three tokens: the last attempt in C and the middle two attempts in D. Accepting up to 8 cm throughout produces 15 tokens: every attempt except A’s 9 cm response. The gradual protocol produces 11.

These totals describe three rules applied to the same fixed record. They do not show what responses would have occurred if a learner actually experienced the alternative rules. A larger token total is not automatically better learning. Keeping the initial criterion forever also fails to distinguish a 7 cm approximation from the final 1 cm target.

Understand: distinguish the mechanism and the evidence

Drawing a destination, pointing toward it and checking the map might form a sequence of different actions. Teaching a sequence is chaining; narrowing acceptable error in the same pointing response illustrates shaping. A visual prompt supplies help before a response; removing that help tests a different issue from merely reducing token frequency. Identify which change a proposed intervention actually makes before naming its mechanism.

The token is only a programmed consequence in this record. To support a claim that it reinforced accurate pointing, collect evidence of a later change in that defined behavior under a suitable comparison. Practice, feedback, a more familiar map and coaching could also explain improvement. A fictional observation table cannot establish which cause operated.

Design a common assessment using a new map, equal instructions, no coaching and the same accuracy measure for all practice conditions. Record accuracy before assessment feedback, and distinguish immediate performance from a delayed retention test. For an AP response, state the operational definition and one plausible confound. For a college methods note, specify assignment, an appropriate comparison, missing-attempt handling and the limits of the sample. Neither proposal becomes research evidence merely because it is written down.

Apply: repair a new protocol

A fictional volunteer practices delivering a two-minute audio description at a self-chosen, comfortable pace. The target is to include three specified landmarks, not to suppress an accent or manner of speaking. Stage 1 accepts at least one landmark; stage 2 accepts at least two; stage 3 accepts all three. Advance after two consecutive qualifying attempts. Counts on the first five attempts are 1, 0, 1, 1, 2. The stage changes only after the attempt that meets the advancement rule.

Write the active stage, token decision and current success streak for each attempt. A reviewer claims that attempt 3 should already use stage 2 because attempts 1 and 3 both qualified. Repair that claim. Then explain why counting landmarks alone cannot establish that the description is accurate or useful to a listener.

Review the unfamiliar shaping protocol

All of attempts 1–4 use stage 1. Their token decisions are yes, no, yes, yes; their consecutive-success counts are 1, 0, 1, 2. Only after attempt 4 does the criterion advance. Attempt 5 uses stage 2, earns a token and starts a new streak of 1. It does not yet advance to stage 3. The full record earns four tokens.

The failed second attempt breaks the streak, so attempts 1 and 3 are not consecutive successes. Landmark count is a narrow measure: an account can mention all landmarks but misdescribe them. Add an independently checked factual-accuracy measure and listener comprehension evidence rather than treating the count as the entire goal.

Keep this activity a paper or notebook audit. An ethical application would require a meaningful, agreed goal, voluntary participation, accessible alternatives and evaluation of unintended effects. Do not make essential services, grades or basic needs conditional on earning these fictional tokens. Improving a measured response is not sufficient justification for any behavior-change program.

The linked OpenStax shaping explanation supports the successive-approximation concept. The museum, criteria, advancement rules and audio-description records are original teaching constructions, not published intervention results.