Token economy · Atlas

1. Introduction

Version followed: the ward-wide token economy as originated by Ayllon and Azrin at Anna State Hospital (1961 onward; published 1965 and in the 1968 book The Token Economy), organised using the standard components laid out by Kazdin (1977), with the levels and social-learning extension of Paul and Lentz (1977) and the patient-rights constraints that followed Wyatt v. Stickney (1971–72). Classroom and home variants are noted where they differ. [uncertain: the Ayllon and Azrin primary texts and Kazdin 1977/1982 are not drawn on in full; their content here comes from abstracts, Wexler 1973, Betgem 1982, Doll et al. 2013 and other secondary accounts.]

A token economy is not a talking therapy. It is a whole-environment motivational system (a ward, class, residential unit or household) in which clearly defined target behaviours earn tokens immediately, and tokens are later exchanged for back-up reinforcers such as goods, privileges and activities. Its account of change is operant: behaviour that is reliably followed by valued consequences becomes more frequent; tokens bridge the delay between behaviour and back-up reward and, because they buy many things, they stay effective whatever the person happens to want at the moment. Tokens are always paired with social praise so that, over time, ordinary social and natural consequences can take over. [source: Wikipedia, Token economy; Ayllon and Azrin 1968 via Liberman 2000; Kazdin 1977 via Doll et al. 2013; Psychology of Learning (TxWes) 13.2]

For a practising therapist this is a programme you design, train others to run, and supervise, not a set of in-session techniques. In outpatient work the same procedures are most often used as a home or classroom token system run by parents or teachers with the therapist as consultant. [synthesis]

2. Framework

The sources describe components and decision points rather than numbered phases; the phase sequence below is my organisation of them. [synthesis, built from Kazdin 1977 via Doll et al. 2013; Cooper, Heron and Heward 2020 via OCALI; Athabasca University token economy pages; Betgem 1982]

Format and dose. Delivered continuously, all day, by all staff in the setting, not in sessions. Inpatient programmes ran for months to years; Paul and Lentz's trial ran over several years and effects in chronic psychosis can take months to appear. Exchange ("store") periods and brief individual review meetings are the only session-like elements; one Dutch long-stay ward held two daily review slots of roughly ten minutes per patient where tokens were paid out with praise. [source: Wikipedia; Betgem 1982] [uncertain: no fixed session count exists in any source]

Phase 0. Assessment, feasibility and consent. Goals: decide whether a token system fits (setting control, staff numbers, client capacity to learn contingencies), obtain consent or advocacy approval, and record baseline frequency of each candidate target behaviour. Move on when baseline data exist for every target and consent/approval is documented. [source: Athabasca, Other issues (baseline); Perlin 1976 (consent as first step); psychiatryonline 1999 unit report (voluntary enrolment)]

Phase 1. Design. Goals: define target behaviours, choose tokens, identify back-up reinforcers, set prices and earning rates, decide on response cost, levels, individual add-ons and data system; write a staff manual and client rule sheet. Move on when the written economy balances (an average client can earn something for minimal effort but more for more) and every rule is written. [source: Doll et al. 2013; Martin and Pear 1998 via Athabasca; Wikipedia]

Phase 2. Staff training and dry run. Goals: all shifts deliver tokens identically. Train with the manual, role play and on-unit coaching; run a trial week in which tokens are earned but mistakes carry little cost, or field-test privately by recording when tokens would have been earned. Move on when staff agree on definitions and, in the field-test version, the client would earn on about 75–80% of opportunities over three days. [source: psychiatryonline 1999 unit report; Cooper et al. 2020 via OCALI]

Phase 3. Orientation and token conditioning. Goals: clients know the rules and tokens acquire value. Explain rules at each person's level, post the earning and price lists, sample the reinforcers, and exchange frequently at first. Move on when clients spend tokens spontaneously and target rates rise above baseline. [source: Kazdin 1977 via Athabasca and Doll et al. 2013; Ayllon et al. 1968 reinforcer sampling]

Phase 4. Active programme and shaping. Goals: build target behaviours from simple self-care and attendance toward social, vocational and independent-living skills; add individual contingencies for non-responders. Review data at least weekly against baseline; advance requirements only when the data say so, not on impression. [source: Betgem 1982 (weekly individual review, graded difficulty); Athabasca, Other issues]

Phase 5. Thinning and levels. Goals: reduce dependence on tokens. Make exchanges less frequent and less predictable, require more behaviour per token, move clients up levels with more privileges and fewer tokens, and pair everything with praise. Move up when criteria at the current level are met over a set period (the period is programme-defined). [source: Kazdin 1977 and McLaughlin and Malaby via Doll et al. 2013; Martin and Pear 1998 via Athabasca]

Phase 6. Transition and maintenance. Goals: behaviour holds under natural consequences. Shift to self-monitoring, social and natural reinforcers, and arrange supportive conditions in the next setting (family, residence, work). Some clients will need an indefinite prosthetic system, which is a legitimate endpoint. [source: Wikipedia (prosthesis framing); Wolf et al. 1987 via Athabasca; Doll et al. 2013]

3. Guiding principles

Define behaviour so it needs no interpretation. If two staff could disagree on whether it happened, rewrite the definition. [source: Kazdin 1977 via Athabasca; Achievement Place reading example]
Choose behaviours that matter in the person's life. Target what will keep being useful and reinforced after the programme, not what is convenient for the institution. [source: Wyatt standards via Perlin 1976; Athabasca rules page] [uncertain: Ayllon and Azrin are often credited with a named relevance-of-behaviour rule; wording not verified]
Reinforce, don't punish. Most of the programme is earning; response cost is a limited, pre-specified supplement. [source: div12 SCP entry; Doll et al. 2013]
Immediacy. Deliver the token as soon as the behaviour occurs; the token exists to make delay bearable. [source: Wikipedia; Doll et al. 2013]
Always pair tokens with specific praise. Praise is what will remain when tokens go. [source: Wikipedia; Betgem 1982]
Use many back-up reinforcers. Tokens that buy many things resist satiation and work across moods and needs. [source: Kazdin 1977 via Doll et al. 2013; TxWes 13.2]
Find reinforcers by watching and asking. What a person often does when free to choose, and what they request, are the best candidates. [source: Wexler 1973 summarising Ayllon and Azrin and the Premack principle]
Basic rights are never earned. Food, bed, privacy, clothing, visitors, religious practice, exercise and dignity are given unconditionally; tokens buy extras above that floor. [source: Wexler 1973 on Wyatt; Betgem 1982]
Participation is voluntary and consent is the first step. In institutions consent is fragile, so document it carefully and allow withdrawal. [source: Perlin 1976; Betgem 1982; psychiatryonline 1999 unit report]
Rules are public and not arbitrary. Clients and staff both know exactly what earns and costs what; staff pay earned tokens even when annoyed with the client. [source: Athabasca rules; Wikipedia]
Consistency across every staff member and shift. Inconsistency is the commonest reason programmes fail. [source: Wikipedia; Kazdin 1982 abstract]
The programme belongs to the team, not one person. Spread knowledge so it survives staff turnover. [source: Betgem 1982]
Decide by data. Measure before and during; advance, hold or change according to the numbers. [source: Athabasca, Other issues]
Shape. Start with requirements the person can already nearly meet and raise them in small steps. [source: Wikipedia; Cooper et al. 2020 via OCALI]
Individualise within the group system. A ward-wide economy plus individual targets for each person. [source: Wikipedia; Betgem 1982]
Plan for fading from day one. Know how tokens will be withdrawn and what will replace them. [source: Doll et al. 2013]
Involve clients and families. Clients can help set contingencies; families can sustain or undermine the system. [source: Wikipedia; Betgem 1982]
Least restrictive, least intrusive. Use the mildest procedure likely to work and expect external scrutiny. [source: Perlin 1976]

4. Intervention inventory

The inventory lists the procedural components of a token economy as the sources describe them. Steps are written in my own words; example language is illustrative throughout. Where an item comes from a later or adjacent source rather than Ayllon and Azrin, it is named.

4.1 Baseline observation

Purpose / when: Before any contingency is introduced, and again whenever a target is added. Shows the size of the problem and gives the yardstick for every later decision. [source: Athabasca, Other issues; Ayllon and Azrin 1965 used direct measurement of ward behaviour, per TxWes 13.2]
Steps: (1) List candidate behaviours. (2) Write a working definition for each. (3) Choose a recording method that fits: frequency counts for discrete acts (showering, attending a group), task completion checklists for chores, time sampling for ongoing states (for example, appropriate social interaction on the ward). (4) Record for long enough to see a stable pattern across days and shifts. (5) Check agreement by having two staff record the same period independently; revise definitions where they disagree. (6) Graph the data.
Example language (illustrative): To staff: "For the next week we only watch and record; nobody gets tokens yet. Tick the sheet each time Mr B is dressed by 8:30, and note anything that made it hard to decide."
Watch for: Definitions that produce low agreement; behaviours that are already frequent (no need to target); very rare problem behaviours (put effort into rewarding the alternative instead). [source: Athabasca, Other issues]

4.2 Target behaviour selection and definition

Purpose / when: Phase 1 and every individual review. Determines what the economy pays for.
Steps: (1) Prioritise behaviours that matter for the person's functioning and future setting: self-care, attending activities, ward or household tasks, work skills, social behaviour, treatment participation, academic work. (2) Exclude anything serving only institutional convenience. (3) Write each as an observable action with a completion criterion (what, when, to what standard). (4) Order targets from easy to harder and start with ones the person can nearly do. (5) Check that prerequisite skills exist; if not, teach them first. (6) Involve the client in choosing targets where possible. (7) Publish the list with token values.
Example language (illustrative): To a client: "Bed made means sheet and blanket pulled up and pillow at the top, by the time breakfast starts. That earns two tokens."
Watch for: Vague targets such as "behaving well" or "reading" (the Achievement Place example: staff paid for comics, parents only for schoolbooks, and the client became confused and angry). Unpaid institutional labour is a legal problem; see 4.20. [source: Athabasca rules page; Wyatt standards via Perlin 1976; Cooper et al. 2020 via OCALI; div12 SCP; Betgem 1982]

4.3 Identifying back-up reinforcers (observation, asking, Premack)

Purpose / when: Phase 1 and whenever spending drops. Tokens are only as strong as what they buy.
Steps: (1) Observe what each person does when free to choose (where they spend time, what they seek, what they hoard). Activities that occur often can be offered as reinforcers for less frequent ones (the Premack principle). (2) Ask clients directly what they would like; catalogue items work. (3) Build a menu across categories: material goods (commissary, snacks, toiletries, clothing), activities (outings, films, TV choice, games), privileges (grounds passes, room choice, extra phone calls, time with a chosen staff member), and services. (4) Include free and low-cost options. (5) Exclude anything that is a right (4.20) or that would disrupt the setting. (6) Review the menu regularly and add idiosyncratic items for individuals.
Example language (illustrative): "If you had an hour and could do anything here on the unit, what would it be? What would you buy if the shop had anything you wanted?"
Watch for: Long-stay or very withdrawn clients may show little interest in anything; that is the cue for reinforcer sampling (4.4). Wexler notes that Ayllon and Azrin's original menu (private screens, a bedroom with a door, ground privileges, religious services, a personal chair, TV) would largely be unlawful to make contingent today. [source: Wexler 1973; Liberman 2000; Betgem 1982; Doll et al. 2013]

4.4 Reinforcer sampling

Purpose / when: When clients rarely buy available reinforcers, at orientation, when introducing a new reinforcer, or when the menu seems to have lost its pull. Especially useful when reinforcers are used infrequently. [source: Ayllon et al. 1968, JABA, via Behaviorist Book Club summary and Wexler 1973]
Steps: (1) Pick an available reinforcer event (a walk, a film, a music session). (2) At the time it is offered, bring everyone to it briefly and free of charge, for a short sample rather than the whole event. (3) Afterwards, offer the full event at its usual token price. (4) Track how many people buy it and for how long. (5) Repeat sampling for reinforcers that are still under-used.
Example language (illustrative): "The music group is starting. Come and sit in for five minutes, no tokens needed, and see whether it is something you'd want to spend on next time."
Watch for: Rising purchase of the sampled reinforcer. In the original reversal study use rose under sampling and stayed somewhat raised after it stopped. Keep samples brief and voluntary in feel; compulsory attendance at a reinforcer conflicts with current rights standards. [synthesis on the voluntariness point]

4.5 Choosing and making tokens

Purpose / when: Phase 1.
Steps: (1) Choose a form suited to age and setting: tangible (chips, tickets, coins, stickers, token board) or recorded (points, checkmarks on a card). (2) Make tokens hard to copy or steal: signed, numbered, or issued in the person's name. (3) Keep them portable so staff can deliver immediately. (4) Decide how clients store them (wallet, folder, jar, point card). (5) For children with strong interests, themed tokens can add value.
Example language (illustrative): "These cards are signed by staff and have your name on them, so only you can spend them."
Watch for: Counterfeiting, theft, bartering between clients; tokens too babyish for adults. [source: Doll et al. 2013; Cooper et al. 2020 via OCALI; Betgem 1982 (named booklets)]

4.6 Establishing tokens as conditioned reinforcers

Purpose / when: Phase 3, or when tokens show no effect. A token has no value until it reliably buys things.
Steps: (1) In the first days, give tokens for very easy behaviours and let clients exchange them almost immediately. (2) Say what each token was for and what it can buy as you hand it over. (3) For clients with intellectual disability or young children, exchange after each token or after a few, then gradually lengthen the interval. (4) Once tokens are clearly valued, move to scheduled exchange (4.9).
Example language (illustrative): "Here's a token for coming to breakfast on time. You can swap it right now for a coffee from the trolley, or keep it toward something bigger."
Watch for: Clients spending tokens as soon as they get them is expected early; failure to spend at all means tokens have not acquired value. [source: Kazdin 1977 via Doll et al. 2013; O'Leary and Drabman 1971 via Doll et al. 2013]

4.7 Pricing and managing the economy

Purpose / when: Phase 1, then weekly. Keeps tokens valuable and earning worthwhile.
Steps: (1) Price back-up reinforcers by desirability and cost: cheap items cheap, highly desired items expensive. (2) Set earning values so that minimal effort buys something but more effort clearly buys more. (3) Start with low prices and raise them, or add luxury items, as skills and earnings grow. (4) Post a price list. (5) Monitor total tokens earned versus spent and the size of savings across the population. (6) Adjust prices or earning rates when hoarding (inflation) or bankruptcy appears. (7) Announce price changes in advance.
Example language (illustrative): To a client: "Prices on the board go up a little on the first of the month; the tokens you already have keep their value until then."
Watch for: Large savings and falling performance (tokens too easy, prices too low); clients unable to afford anything (too hard, too expensive). [source: Wikipedia; Cooper et al. 2020 via OCALI; Doll et al. 2013] [synthesis: earned-versus-spent monitoring as a routine check]

4.8 Token delivery

Purpose / when: Every time a target behaviour occurs.
Steps: (1) Deliver as soon as possible after the behaviour. (2) Name the behaviour specifically. (3) Add genuine praise and eye contact. (4) Record the delivery. (5) Deliver earned tokens regardless of how the client behaved earlier or how staff feel about them. (6) Do not negotiate or argue at delivery.
Example language (illustrative): "You finished the laundry and folded it all; that's three tokens. Nice job getting it done before lunch."
Watch for: Tokens delivered silently (no social pairing), late, or only by some staff; staff withholding earned tokens as a punishment. [source: Wikipedia; Doll et al. 2013; Betgem 1982]

4.9 Exchange periods (the store)

Purpose / when: Throughout. Exchange gives tokens their meaning.
Steps: (1) Fix where and when exchange happens: immediate, at set times (end of class, end of day, at transitions), or at variable times. (2) Display the menu and prices. (3) Start with frequent exchange, especially for lower-functioning clients and children with attention problems. (4) Reduce store hours as clients meet expectations. (5) If someone has too few tokens, respond briefly and supportively without lecturing.
Example language (illustrative): "You're two short for the outing today. You can keep those and add to them tomorrow; the next outing is Thursday."
Watch for: Variable exchange times were associated with higher response rates and better maintenance in classroom studies. [source: Doll et al. 2013 citing McLaughlin and Malaby; Cooper et al. 2020 via OCALI]

4.10 Shaping and prompting

Purpose / when: When the full target is beyond the person's current repertoire.
Steps: (1) Break the target into steps or set a lower initial criterion. (2) Pay tokens for the first approximation. (3) Raise the criterion when the step is reliable on the data. (4) Use prompts (reminders, instructions, modelling) as antecedents and fade them as the behaviour establishes.
Example language (illustrative): "This week, a token for five minutes in the group with your phone away. Next week we'll aim for ten."
Watch for: Raising criteria on impression rather than data, which risks failure and loss of interest. [source: Wikipedia (5- to 30-minute attention example); Liberman 2000 (prompts and consequences); Athabasca, Other issues]

4.11 Individual contingencies within the group economy

Purpose / when: For clients whose needs the ward-wide list does not cover, or who are not responding.
Steps: (1) Meet the client and write an individual programme with one to three personal targets (for example hygiene, eating, cursing, self-harm alternatives) and their token values. (2) Set difficulty so the target is neither too hard nor too easy. (3) Review weekly with the client and adjust. (4) For a problem behaviour, either pay bonus tokens for its absence or for an alternative behaviour, or (sparingly) attach a cost. (5) Record the plan where all shifts see it.
Example language (illustrative): "Let's pick one thing that would make the biggest difference for you this week. How about getting to the morning meeting? That will earn you four extra tokens a day on your own plan."
Watch for: Plans that drift without review; plans that become disguised punishment. [source: Betgem 1982 (individual programmes reviewed every seven days); psychiatryonline 1999 unit report; Wikipedia]

4.12 Group contingencies

Purpose / when: Classrooms and units where tracking individuals is impractical, or to harness peer support.
Steps: (1) Define a criterion for the whole group (for example, the class completes the task with no more than a set number of reminders). (2) Decide whether meeting it earns a group reward or individual rewards. (3) Post the group's progress. (4) Combine with individual tokens where some members need more.
Example language (illustrative): "If the room is tidy by 3:00, everyone gets ten minutes of free time."
Watch for: Peer pressure on the member who keeps the group from earning; drop or individualise if scapegoating appears. Betgem's ward trial of group-level programmes failed through staffing and timing constraints. [source: Doll et al. 2013 (group-oriented systems easier to run); Betgem 1982] [synthesis: scapegoating check]

4.13 Response cost (fines)

Purpose / when: Optional, for specified problem behaviours that rewarding alternatives alone has not reduced, or for serious safety violations.
Steps: (1) List in advance each behaviour that costs tokens and how many. (2) Keep costs proportionate and fewer than earning opportunities. (3) Take the fine calmly, name the rule, and do not add a lecture. (4) Keep reinforcing the incompatible desirable behaviour. (5) Never let a client go into token debt so deep that earning seems pointless; cap fines. (6) Review response-cost frequency in staff meetings.
Example language (illustrative): "Smoking in the bedroom is on the safety list, so that's the fine we agreed. You can start earning again straight away at the next group."
Watch for: Staff becoming more negative when response cost is in use; the economy turning into a punishment system. One acute unit limited costs to serious safety violations (for example smoking in bedrooms or acts that would involve police outside) while letting restricted patients keep earning. [source: Doll et al. 2013 citing Iwata and Bailey 1974; psychiatryonline 1999 unit report; Wikipedia] [synthesis: debt cap]

4.14 Level systems

Purpose / when: To structure progression from dense external reinforcement toward more autonomy, and to link privileges to safety or skill.
Steps: (1) Define levels from entry to highest, each with entry criteria, earning rules and the privileges available. (2) Higher levels carry more freedom and access, fewer token-checked behaviours, and longer intervals between exchange. (3) Specify how and when a client moves up or down. (4) Review level status on a fixed schedule. (5) Keep basic rights at every level.
Example language (illustrative): "You've met the level 2 goals every day for two weeks, so from tomorrow you're on level 3. That means grounds passes and your points are checked once a day instead of every shift."
Watch for: Levels used to restrict rights; clients stuck on entry level. Paul and Lentz's social-learning programme used a step system in which residents progressed toward more self-direction before discharge. [source: Doll et al. 2013 (Filcheck et al. 2004 six-level preschool system); psychiatryonline 1999 unit report] [uncertain: Paul and Lentz level details not verified against the book]

4.15 Lotteries and bonus systems

Purpose / when: To stretch a small reward budget or add excitement.
Steps: (1) Target behaviours earn tickets. (2) At the end of the period, draw tickets for back-up rewards. (3) Keep a guaranteed small reward alongside the lottery so effort is never wholly unrewarded.
Example language (illustrative): "Every finished worksheet puts a ticket in the jar; on Friday we draw three for the prize box."
Watch for: Clients who never win losing interest. [source: Doll et al. 2013 citing McLaughlin and Williams 1988] [synthesis: guaranteed small reward]

4.16 Saving and banking

Purpose / when: Once tokens are established, to build planning and delay tolerance.
Steps: (1) Offer some high-priced items that require saving across days or weeks. (2) Provide a safe way to store or record savings. (3) Help the client set a savings goal and track it. (4) Watch the economy for excessive hoarding (4.7).
Example language (illustrative): "The trip to town costs 60. You've got 25. If you keep earning like this week, you'll have it by the 20th."
Watch for: Saving with no spending at all, which usually means current targets have become too easy to need effort. [source: Wikipedia (forward planning and saving)]

4.17 Rule orientation for clients

Purpose / when: At admission or programme start and whenever rules change.
Steps: (1) Explain what earns, what costs, prices and exchange times, in language fitted to the person's ability. (2) Give a written copy to clients who can use one and post the lists. (3) Show that earning does not depend on staff mood. (4) Explain rights that are not part of the economy and how to withdraw or complain. (5) Remind new clients to collect and spend tokens in the first days.
Example language (illustrative): "This sheet lists everything that earns points and what you can buy with them. Meals, your room, visits and phone calls to your lawyer are yours no matter what. Joining is your choice, and you can stop."
Watch for: Clients who seem confused about why they did or didn't earn; this points to unclear definitions. [source: Kazdin 1977, 1985 via Athabasca; psychiatryonline 1999 unit report; Betgem 1982]

4.18 Client involvement, self-monitoring and self-administration

Purpose / when: From design onward, increasingly in later phases; supports dignity and generalisation.
Steps: (1) Invite clients to suggest targets, reinforcers and prices. (2) Write individual plans with the client. (3) In later phases have clients record their own target behaviour and compare it with staff records, paying a bonus for accuracy. (4) Progress toward clients awarding their own points under spot checks.
Example language (illustrative): "From this week you mark your own card each evening. If your card matches mine, you get a bonus point."
Watch for: Inflated self-ratings; reduce responsibility if accuracy falls. [source: Wikipedia (clients help set contingencies); Turkewitz et al. 1975 via Doll et al. 2013 (self-monitoring for generalisation); Zlomke and Zlomke 2003 listed in Wikipedia] [uncertain: accuracy-matching bonus is a common classroom variant, not verified against a primary source here]

4.19 Staff manual, training and integrity monitoring

Purpose / when: Phase 2 and continuously. Consistency is what makes a token economy work.
Steps: (1) Write a staff guide: which behaviours are reinforced and how, how tokens and back-ups are delivered, when reinforcement is available, what data to record and how, and each person's duties. (2) Run classroom training with manual review, questions and role play. (3) Coach on the unit during a trial period. (4) Name a coordinator, but spread knowledge so the programme does not rest on one person. (5) Brief ancillary staff (stores, finance, cleaners) and families. (6) Check integrity by spot observation of delivery and by comparing records across shifts; give feedback. (7) Retrain rotating and new staff with a short standardised package.
Example language (illustrative): To staff: "In the role play, you're on nights and Ms D asks for her bedtime token after an argument earlier. Show us what you do."
Watch for: Shift-to-shift drift, staff disagreeing with the philosophy, reliance on a single champion. [source: Martin and Pear 1998 via Athabasca; psychiatryonline 1999 unit report; Betgem 1982; Kazdin 1982 abstract]

4.20 Rights safeguards, consent and review

Purpose / when: Before launch and continuously. Protects clients and keeps the programme lawful.
Steps: (1) Write a list of non-contingent entitlements: adequate meals, bed, privacy, own clothing, visitors, religious practice, outdoor exercise, communication, basic comfort. Tokens buy only extras above this floor. (2) Obtain informed consent; in institutions document capacity and voluntariness with special care, and allow withdrawal. (3) Do not make institutional maintenance labour compulsory; work that benefits the institution may need to be voluntary and paid. (4) Have a physician rule out correctable physical causes before targeting a problem behaviour, as Wyatt standards required for developmental disability settings. (5) Submit the programme to an independent human-rights or ethics committee; set time limits and review dates. (6) Choose the least restrictive effective procedure.
Example language (illustrative): "Before you decide: here is what the programme pays for and what it never touches. You can say no now or stop later without losing anything you're entitled to."
Watch for: Any item on the menu that is arguably a right; advocacy concerns about punitiveness or lack of individualisation (one unit answered these by making the programme voluntary and adding individual plans). [source: Wexler 1973; Perlin 1976; psychiatryonline 1999 unit report; Betgem 1982]

4.21 Data review and contingency check

Purpose / when: Weekly and whenever results are unclear. Ayllon and Azrin showed that behaviour rose when tokens were contingent and fell when the same tokens were given non-contingently, which is the logic behind checking that the contingency, not attention or novelty, is doing the work.
Steps: (1) Graph each target against baseline. (2) Decide: advance, hold, or change the programme. (3) If a target is not changing, check delivery (is it immediate and consistent?), exchange (is it frequent enough?), and back-ups (are they still wanted?). (4) Change one element at a time and observe.
Example language (illustrative): To the team: "Showering hasn't moved in two weeks. Nights are recording it but days aren't paying for it; let's fix that before we change anything else."
Watch for: Decisions based on staff impression; never run deliberate non-contingent phases with clients for demonstration outside an approved research protocol. [source: Ayllon and Azrin 1965/1968 via TxWes 13.2; Doll et al. 2013; Athabasca, Other issues] [synthesis: the research-protocol caution]

4.22 Thinning, fading and transfer to natural reinforcers

Purpose / when: Phases 5–6, planned from the start.
Steps: Use one or more, gradually: (1) require more behaviour per token; (2) restrict which behaviours still earn tokens; (3) lengthen the delay between behaviour and token, or between token and exchange; (4) make exchange times less predictable; (5) raise back-up prices; (6) keep or increase praise and natural rewards (free time, a break, privileges ordinary life provides); (7) add self-monitoring; (8) arrange support in the next environment (family, residence, workplace) so the behaviour keeps being reinforced; (9) accept a continuing prosthetic system where needed.
Example language (illustrative): "You've been on time every day for a month, so we're stopping points for that. I'll still notice, and the late-morning coffee break stays yours."
Watch for: Collapse of behaviour after withdrawal means fading was too fast or the new environment does not support it; step back one stage. [source: Martin and Pear 1998 and Kazdin 1977 ch. 7 via Athabasca; Doll et al. 2013; Wolf et al. 1987 via Athabasca; Wikipedia]

4.23 Outpatient adaptation: home or classroom token systems

Purpose / when: When a therapist sees a child, an adult with intellectual disability or a family and wants a contingency system in the home or school. [synthesis: the sources describe classroom and single-person systems; the consultation format is my framing]
Steps: (1) With parents or teachers, choose two or three observable targets. (2) Pick a token board, chart or points. (3) Build a reinforcer menu with the child. (4) Practise delivery and praise in session by role play. (5) Begin with frequent exchange. (6) Review the chart at each session, troubleshoot using 4.21, and plan fading.
Example language (illustrative): To a parent: "When she gets dressed before the timer goes, hand her the star straight away and say exactly what she did. Let's practise that now."
Watch for: Parents using tokens inconsistently, taking away earned stars in anger, or adding new targets too fast. [source: Wikipedia (single-person designs, family role); Doll et al. 2013] [synthesis]

5. Handling problems

Non-responders (stuck clients). Expect a minority of clients, often cited as roughly 5–20% (about 15% in Kazdin's early estimate), to show little or no change. Response: check delivery, exchange and back-ups (4.21); sample reinforcers (4.4); find idiosyncratic reinforcers by observation and asking (4.3); lower the criterion and shape (4.10); write an individual plan (4.11); involve family. If still flat, use another intervention; severe behaviour may need other approaches. [source: Wikipedia; Betgem 1982 citing Kazdin 1973; Doll et al. 2013; Wexler 1973]
Poor candidates. Severe amnestic or dementing conditions (impaired learning of contingencies), severe depression, and highly educated adults who find the system childish were noted as poor fits on one long-stay ward. Response: screen for these, adapt (adult-styled points, self-administration, contract format) or choose another treatment. [source: Betgem 1982] [synthesis: adaptations]
Client resistance and feelings of humiliation. Adults may experience tokens as demeaning. Response: make participation voluntary; involve the person in choosing targets and rewards; use adult forms (points, credits, pay-like records); frame the system honestly as an earning structure; meet resistance with engagement and encouragement, not confrontation. [source: Wikipedia; Athabasca, Other issues (Spiegler and Guevremont 1998); Betgem 1982]
Refusal or withdrawal. Response: accept it; basic rights and other treatment continue unchanged; offer to revisit later. Coercion here is both an ethical and a legal problem. [source: Betgem 1982; Perlin 1976; psychiatryonline 1999 unit report]
Non-adherence by staff (the most common failure). Inconsistent delivery, withholding earned tokens, ad hoc rule changes. Response: written manual, role-play training, on-unit coaching, integrity spot checks, regular staff meetings with data, a coordinator plus shared ownership; brief new and rotating staff with a standard package. [source: Wikipedia; Kazdin 1982 abstract; Betgem 1982; psychiatryonline 1999 unit report]
Staff philosophical opposition. Clinicians trained in other models may resist. Response: involve them in design, show the baseline and outcome graphs, and keep the programme reinforcement-led rather than punitive. [source: Liberman 2000 (clinician resistance as a historical barrier)] [synthesis: responses]
Punitive drift. Response cost expands, staff become more negative, the economy becomes a discipline system. Response: audit the ratio of tokens paid to tokens fined; restrict costs to a short pre-agreed list; refocus training on catching good behaviour. [source: Doll et al. 2013 (Iwata and Bailey 1974)] [synthesis: ratio audit]
Escalation or aggression around fines or denied purchases. Response: deliver fines calmly and without debate; never physically retrieve tokens; let the person keep earning immediately; handle dangerous behaviour through the setting's usual safety procedures, not the economy. [synthesis]
Overwhelm or confusion. Too many targets or rules for the person's capacity. Response: cut to one or two targets, use more tangible tokens and immediate exchange, repeat orientation (4.17). [source: Doll et al. 2013 (frequent exchange for lower-functioning participants)] [synthesis: reduce targets]
Satiation and loss of value. Response: widen and rotate the menu, add new items, re-sample. [source: Doll et al. 2013; Kazdin 1977 via Doll et al.]
Inflation, hoarding or bankruptcy. Response: adjust prices and earning values (4.7), add saving goals, announce changes in advance. [source: Wikipedia] [synthesis: economic adjustments]
Theft, counterfeiting, bartering, sabotage. Response: named or signed tokens, secure storage, point records rather than objects; on one ward fraud led to closer monitoring rather than repayment demands. [source: Doll et al. 2013; Betgem 1982]
Loss of gains after the programme (generalisation failure). Response: plan fading from the start (4.22); build self-monitoring; arrange supportive conditions after discharge; accept a long-term prosthetic arrangement where that is what keeps the person functioning. [source: Athabasca, Other issues (Wolf et al. 1987); Wikipedia]
Concern about intrinsic motivation and "bribery". Response: target behaviours the person is not currently doing (so there is little intrinsic motivation to undermine), pair with praise, fade tokens, and explain the rationale to teachers, parents and community. [source: Doll et al. 2013; Wikipedia] [synthesis: targeting rationale]
Families undermining the system. Response: include families in orientation, give them a role (for example, home visits as earned privileges where appropriate), keep them informed; one ward had families on an oversight board for finances. [source: Wikipedia; Betgem 1982]
Rights conflicts. A proposed reinforcer is or may be an entitlement. Response: remove it from the menu and give it non-contingently; use idiosyncratic, Premack-type or sampled reinforcers instead; take doubtful items to the ethics or rights committee. [source: Wexler 1973; Perlin 1976]
Ruptures in the therapeutic relationship. When a client feels treated unfairly over tokens. Response: review the record together, correct staff errors openly and pay what was owed, acknowledge the frustration, and if the rule itself is unfair change it with the client. Keep the clinician's individual role separate from day-to-day fining where possible so the therapeutic relationship is not tied to enforcement. [synthesis]
Institutional and time pressure. Gains can take months, conflicting with short stays and funding demands. Response: set realistic review horizons, use the data to show progress, and design shorter, simpler systems for brief admissions. [source: Wikipedia; div12 SCP (adaptable to shorter programmes)] [synthesis: design response]
Trauma histories and control. Highly controlled environments can echo past coercion. Response: maximise choice, voluntariness and transparency; avoid deprivation-based reinforcers; consult the person about anything that feels controlling. [synthesis; not addressed in the sources]

Sources

Sources: