Differential Reinforcement – Quick Review
Page 1
- Key terms: DRA, DRO, DRI, DRL, FI-DRO, VI-DRO, FM-DRO, VM-DRO
- Differential reinforcement = reinforce one response class, withhold for another
- Illustrative anecdote: Prince Albert uses DRA to reinforce Queen Victoria’s calm request
- Rationale: Prefer DR over extinction/punishment due to fewer side-effects
Page 2
- Differential reinforcement: (a) reinforce other/alternative behavior or reduced rate, (b) withhold for problem behavior
- Forms most used for reduction: DRA, DRO, DRL
- DRA = reinforce desired alternative, place problem on extinction/lean schedule
- Matching law: response allocation favors denser reinforcement schedule
Page 3
- DRA example (Fernando): attending to appropriate speech, ignoring inappropriate -> inappropriate ↓, appropriate ↑
- Alternative & problem behaviors often mutually exclusive; if incompatible ⇒ termed DRI
- DNRA: use escape (negative reinforcement) for alternative behavior when problem maintained by escape
Page 4
- DRA guideline highlights:
• Choose alternative already in repertoire, easier/no harder than problem, occurs often, socially reinforced
• Use potent, function-based reinforcers; quality > magnitude consistency
• Start with continuous reinforcement, then thin; withhold all reinforcement for problem (extinction)
Page 5
- Partial fidelity study: some reinforcement errors still allow DRA success if initial full-fidelity exposure given; booster sessions recommended
- Watch for relapse when thinning schedule, changing setting/people
- Resurgence risk; mitigating tactic: reinforce multiple alternative behaviors (serial DRA)
Page 6
- Combine DRA with other procedures (blocking, time-out, stimulus fading, DRO) for severe behaviors
- Example: Instructional fading + DRA reduced task-related aggression more than DRA alone
- Introduces DRO: reinforce absence of problem during interval (interval) or at check moment (momentary)
Page 7
- Four processes possibly driving DRO effects: AO/satiation, extinction, punishment (delay), adventitious reinforcement of other behaviors
- DRO variants = FI-DRO, VI-DRO, FM-DRO, VM-DRO (interval/momentary × fixed/variable)
Page 8
- FI-DRO: set fixed interval; reset on problem; reinforce if none
• Example: 5-min timer for class disruptions → free time - Gradual thinning of interval/session length maintains gains (scratch SIB case study)
Page 9
- VI-DRO: varied intervals average a value; “distance-based” bus example reduced disruptive ride behavior
- Conjunctive schedules: combine DRO with task FR requirement
- Momentary DRO: only absence at interval end matters; easier to implement but risk ineffectiveness
Page 10
- Signaled vs unsignaled FM-DRO: signals can enhance or hinder; subtle therapist cues may act as signals
- Recommendation: use caution; monitor effects individually
Page 11
- VM-DRO may yield: easier monitoring, higher reinforcement rates than FI-DRO
- DRO guidelines:
• Initial interval ≤ mean baseline IRT
• Avoid reinforcing other bad behaviors
• Gradually lengthen interval once stable improvements
• Extend to natural settings; often combine with other procedures
Page 12
- Limitations: possible inadvertent reinforcement of other inappropriate behaviors; very short intervals can be labor-intensive
- DRL introduced: decrease (but not eliminate) high-rate behaviors by reinforcing low rates or longer IRTs
Page 13
- DRL variants:
• Full-session: reinforce if total responses ≤ criterion
• Interval: divide session; reinforce if responses ≤ criterion per sub-interval
• Spaced-responding: reinforce a response separated by ≥ specified IRT - Used when behavior acceptable at lower rates (e.g., hand raising)
Page 14
- Full-session DRL example: talk-outs ↓ when ≤3 per session earned play time
- Interval DRL example: 2-min blocks, star if ≤1 misbehavior, stars exchange for recess
- Spaced-responding DRL: used for rapid eating, stereotypy—gradually increase required pause between responses
Page 15
- Selecting DRL type:
• Spaced-responding when behavior must continue at low rate (immediate SR+)
• Full/interval acceptable if rate can drop to zero - Limitations: slow change, not for dangerous behaviors, monitoring effort high
- Baseline guides initial criteria; thin gradually (rules: after 3 sessions meeting criterion or ≥90% reinforcement)
Page 16
- Summary of key points:
• Differential reinforcement = reinforce alternative/other/low-rate behaviors, extinguish problem
• DRA/DRI strengthen alternative; ensure potent SR+, full then thin; watch relapse, combine with others
• DRO reinforces zero/omission; choose proper interval, avoid reinforcing other problem, extend & combine
• DRL slows but retains behavior; choose variant, base criteria on baseline, thin gradually, provide feedback