Acquiring complex behavioral repertoires represents one of the foundational challenges in developmental psychology, neurorehabilitation, and behavioral pedagogy. Backward chaining provides an empirically validated, highly systematic instructional methodology designed to establish intricate multi-step behavioral sequences by prioritizing the final response in a chain. By engineering instructional trials so that the learner encounters immediate, unconditioned or naturally occurring terminal reinforcement upon completing the target step, backward chaining effectively circumvents instructional frustration and optimizes skill acquisition.
Backward Chaining
1. Concise Definition
Backward chaining is an operant conditioning instructional procedure in which a complex behavioral sequence is broken down into discrete steps via task analysis, and instruction begins with the final step of the chain. In this procedure, the trainer performs or completely prompts all initial steps in the sequence until the terminal step is reached, at which point the learner is prompted and reinforced for independently completing that final component. Once mastery of the final step is demonstrated, the criterion shifts backward: the trainer completes all preceding steps up to the second-to-last step, requiring the learner to perform the final two steps sequentially to access the terminal reinforcement.
This progression continues systematically in reverse order until the individual executes the entire chain from the initial discriminative stimulus to the final terminal reinforcer without external prompting. By anchoring instruction to the point in time closest to the delivery of reinforcement, backward chaining establishes a powerful sequence of conditioned reinforcers. Each completed component serves simultaneously as a conditioned reinforcer for the preceding behavior and as a discriminative stimulus for the subsequent behavior.
Beyond its primary clinical and educational applications within applied behavior analysis, the term also designates a goal-driven inference engine strategy in artificial intelligence and formal logic. In automated theorem proving, backward chaining works backward from a proposed goal state or hypothesis to identify the antecedent conditions and inference rules required to satisfy that objective.
2. Etymology & Linguistic Origin
The term backward chaining combines the spatial-directional adverb backward—derived from the Old English bæcweard (composed of bæc, meaning “back,” and the directional suffix -weard, denoting movement toward a specified direction)—with the verbal noun chaining. The word chain traces its linguistic lineage through Middle English chaine and Old French chaeine, ultimately originating from the Latin catena, meaning “a series of linked rings, a fetter, or an uninterrupted series.”
The concept entered the scientific lexicon during the mid-twentieth century through behavioral psychology and operant methodology. B. F. Skinner and early experimental behavior analysts operationalized behavioral sequences as “chains” (or catenations) to denote serially dependent operants. When applied behavior analysts formalized systematic task-analytic methods in the 1960s and 1970s for individuals with severe developmental disabilities, instructional designers explicitly appended the modifiers forward and backward to denote the temporal direction in which instructional prompts and mastery criteria were introduced relative to the task’s natural chronological performance.
3. Pronunciation & Grammatical Form
Pronunciation: /ˈbækwərd ˈtʃeɪnɪŋ/
Grammatical Form: Compound noun phrase (uncountable). It is frequently employed as an attributive modifier or nominal adjunct (e.g., “backward chaining procedure,” “backward chaining paradigm,” “backward chaining instructional protocol”). In computational linguistics and logic programming, it functions identically as a singular noun phrase referring to a deduction algorithm.
Usage Notes: In academic literature, the term should not be hyphenated when functioning as an unhyphenated compound noun, though hyphenation is occasionally observed when used adjectivally before a noun (e.g., “backward-chaining methodology”). The agentive verbal form is typically rendered periphrastically (e.g., “teaching the skill via backward chaining” rather than “backward-chaining the participant”).
4. Detailed Conceptual Explanation
To understand backward chaining, one must first grasp the behavioral architecture of a stimulus-response chain. A behavioral chain is a sequence of discrete behaviors, each associated with a particular stimulus condition, that culminates in a terminal reinforcer. Within this sequence, every intermediate response produces a stimulus change that serves a dual functional role: it acts as a conditioned reinforcer for the response that preceded it and simultaneously functions as a discriminative stimulus (SD) that sets the occasion for the subsequent response. The ultimate integrity of the entire chain is sustained by the terminal reinforcer delivered upon completion of the final link.
In standard forward acquisition, a learner must execute novel, effortful responses while situated at the greatest temporal distance from the terminal reinforcer. This temporal delay frequently leads to instructional extinction, behavioral drift, or high rates of escape-maintained disruptive behaviors, particularly among individuals with severe cognitive impairments or attention-deficit profiles. Backward chaining circumvents this limitation by restructuring the reinforcement landscape. Because the learner begins their active engagement at the very threshold of task completion, the temporal latency between their effortful behavior and the delivery of the primary or established conditioned reinforcer is minimized to near zero.
Consider the logistical execution of backward chaining across successive instructional sessions. The practitioner begins by conducting a meticulous task analysis, decomposing a target behavioral repertoire (such as handwashing, donning a jacket, or preparing an enteral feeding apparatus) into an invariant sequence of operationalized steps. During the initial instructional tier, the practitioner physically guides or performs Steps 1 through n-1. The learner is then presented with the discriminative stimulus corresponding to Step n (the final step). With appropriate prompt fading, the learner executes Step n independently, receiving immediate access to the terminal reinforcer (such as tangible reinforcement, praise, or the natural consummatory reward of task completion).
Once the terminal link meets the predefined criterion for mastery—typically defined as consistent, unprompted correct performance across consecutive trials or sessions—the intervention shifts to the penultimate link (Step n-1). The practitioner prompts or completes Steps 1 through n-2, and the learner executes Step n-1. Crucially, the completion of Step n-1 produces the environmental stimulus that signals Step n, which the learner has already mastered. The learner immediately transitions into performing Step n, followed at once by the terminal reinforcer. This instructional sequence steadily expands backward toward the initial antecedent event, ensuring that the learner continuously practices newly trained behaviors directly into previously mastered chains of established competence.
An essential advantage of backward chaining is its psychological preservation of the natural contextual closure of the task. Rather than experiencing an interrupted or incomplete task routine during early learning trials, the individual consistently experiences the satisfying, functional completion of the activity. This structural feature renders backward chaining particularly effective for self-help, vocational, and adaptive daily living skills, where the functional outcome of the activity provides an intrinsic conditioned reinforcing effect.
5. Historical Development
The theoretical origins of backward chaining trace to the early experimental investigations of operant conditioning conducted by B. F. Skinner in the 1930s and 1940s. Skinner demonstrated that complex non-human animal behaviors—such as a rat pulling a string to release a marble, carrying the marble across a platform, and dropping it down a chute to obtain food—could be constructed through chaining techniques. In his 1938 classic The Behavior of Organisms, Skinner outlined the mechanics of catenated behavioral repertoires and the dual nature of chained stimuli.
During the late 1950s and 1960s, the emergence of early applied behavior analysis saw researchers translating these laboratory findings into human clinical contexts. Psychologists working in institutional settings for individuals with severe intellectual disabilities and psychiatric conditions discovered that conventional verbal instructions and forward shaping frequently failed to establish complex functional living repertoires. Early pioneers such as Teodoro Ayllon, Nathan Azrin, and Montrose Wolf adapted operant methods to address profound skill deficits.
A critical milestone occurred in the late 1960s and 1970s with the refinement of systematic task analysis by researchers such as Marc Gold, who demonstrated through the “Try Another Way” model that individuals with moderate and severe intellectual disabilities could master sophisticated industrial assembly tasks using structured behavioral chaining protocols. During this period, researchers such as Robert Bellamy and associates systematically examined backward chaining in vocational habilitation, documenting that backward progression markedly reduced error rates and accelerated time to independent mastery compared to unstructured trial-and-error learning.
In the 1980s and 1990s, research shifted toward comparative analyses, juxtaposing backward chaining with forward chaining and total-task presentation protocols. Behavior analysts such as Mark Wolery, David Gast, and Maureen Conroy conducted rigorous single-case experimental designs to isolate the instructional variables that dictate when backward chaining is empirically superior. Concurrently, computer science and artificial intelligence adapted the theoretical framework of backward chaining to develop expert inference systems (such as MYCIN), embedding the term within formal computational reasoning models.
6. Theoretical Foundations
The primary theoretical foundation of backward chaining is rooted in radical behaviorism and the experimental analysis of behavior. Specifically, the protocol relies on three interlocked behavioral mechanisms: the conditioned reinforcement hypothesis of chained schedules, the discriminative stimulus control of behavior chains, and stimulus prompt fading paradigms.
The conditioned reinforcement hypothesis explains how intermediate behaviors in a chain are sustained in the absence of an immediate unconditioned reinforcer. According to this framework, an environmental stimulus that reliably precedes a reinforcing event acquires conditioned reinforcing properties through higher-order classical or operant conditioning. In a backward chain, because Step n is directly followed by the terminal reinforcer SR+, the stimulus condition resulting from the completion of Step n-1 acquires reinforcing power. Therefore, when the learner performs Step n-1, the emergence of the environmental state for Step n immediately reinforces Step n-1, while simultaneously serving as an evocative antecedent (SD) for Step n. The chain is held together by this interlocking stimulus architecture: SD1 → R1 → SR+cond / SD2 → R2 → … → Rn → SR+term.
A secondary theoretical foundation originates in cognitive learning theories and instructional design frameworks concerning cognitive load and executive functioning. Learners with neurodevelopmental disorders, acquired brain injuries, or executive dysfunction often exhibit deficits in working memory, working-memory updating, and planning. When instructed via forward chaining, the learner faces an open-ended cognitive horizon: they complete an initial step without a concrete experiential grasp of how that specific motor program links to the overarching objective, leading to elevated cognitive fatigue. In backward chaining, cognitive load is reduced because the terminal goal is immediately proximal, providing an unambiguous perceptual frame of reference that grounds executive processing.
Finally, errorless learning theory intersects significantly with backward chaining protocols. By combining backward chaining with most-to-least prompt hierarchies on newly introduced steps and full physical or visual prompting on all unlearned antecedent steps, practitioners prevent the commission of unreinforced errors. Errorless frameworks posit that the execution of incorrect responses creates competing stimulus-response associations that actively interfere with retention. Backward chaining minimizes error generation by ensuring that independent execution is restricted exclusively to the target step currently under instructional control, while all surrounding links are strictly mediated by the instructor.
7. Key Components, Types & Dimensions
Implementing backward chaining systematically requires a thorough understanding of its constituent structural components and variations:
- Comprehensive Task Analysis: The foundational operationalization of a target behavioral repertoire into discrete, sequentially ordered, observable behavioral units with explicit objective criteria for mastery at each step.
- Standard Backward Chaining: The pure form of the procedure wherein the instructor executes or fully prompts Steps 1 through n-1, teaches Step n to mastery, and then systematically works backward to teach Step n-1, Step n-2, and so forth, delivering the terminal reinforcer solely upon completion of the final step.
- Backward Chaining with Leaps Ahead: An efficiency-oriented variation in which not every single step in the task analysis is systematically trained. The practitioner probes intermediate steps; if the learner demonstrates emergent or incidental acquisition of intermediate links without formal instruction, the instructor “leaps ahead” over those mastered components to train earlier unmastered behaviors, substantially reducing instructional time.
- Prompting Hierarchies: The calibrated assistance protocols applied to the target step, ranging from most-to-least prompting (e.g., full physical guidance fading to partial physical, modeling, gestural, and independent performance) to least-to-most prompting, depending on the learner’s error profile.
- Terminal Reinforcement Schedule: The predefined delivery of primary, secondary, or generalized conditioned reinforcers immediately upon completion of the terminal response of the chain, maintaining the motivational valence of the entire sequence.
- Systematic Mastery Criteria: Quantifiable behavioral thresholds (e.g., 100% independent correct performance across three consecutive sessions with two distinct interventionists) that govern when instruction transitions from step k to step k-1.
8. Examples & Illustrative Cases
The mechanics of backward chaining are best illustrated through concrete clinical and educational scenarios spanning different age groups, developmental levels, and functional domains.
Case 1: Self-Care Acquisition in Pediatric Neurodevelopmental Therapy
A seven-year-old child diagnosed with Level 3 autism spectrum disorder exhibits significant difficulty acquiring independent shoe-tying. The occupational therapist establishes a 5-step task analysis: (1) cross the laces into an “X”; (2) tuck one lace under and pull tight; (3) form the first loop; (4) wrap the free lace around the loop and push it through to create a second loop; (5) grasp both loops and pull them firmly to create a knot.
Utilizing backward chaining, the therapist conducts Steps 1 through 4 entirely for the child. The child is presented with the nearly finished shoes where both loops are formed and positioned. The therapist prompts the child to grasp both loops and pull them tight (Step 5). Upon pulling the loops, the shoe is successfully tied, and the child immediately receives enthusiastic social praise and access to a preferred tablet game. Over subsequent days, once the child independently pulls the loops across five consecutive trials, the therapist completes Steps 1 through 3, requiring the child to wrap the lace and push it through (Step 4) and then pull the loops tight (Step 5). Through this reverse sequence, the child acquires the full repertoire without the behavioral outbursts that had characterized prior forward-chaining attempts.
Case 2: Post-Stroke Neurorehabilitation in Geriatric Occupational Therapy
A 68-year-old individual recovering from a left-hemisphere cerebrovascular accident presents with unilateral hemiparesis and ideational apraxia, struggling to relearn the morning routine of preparing a cup of instant tea. The multi-step task analysis includes: (1) fill kettle with water, (2) turn on kettle, (3) place teabag in mug, (4) pour boiling water into mug, (5) allow to steep and remove teabag, and (6) add a splash of milk and stir.
The clinician performs Steps 1 through 5. The patient is handed the milk jug and spoon to perform Step 6. Completing this final step results in immediate access to the hot, ready-to-drink beverage—the natural, intrinsically reinforcing consummatory outcome. As the patient demonstrates consistent motor control and procedural recall for Step 6, the therapist fades support on Step 5, then Step 4, gradually building the patient’s stamina and self-efficacy while preventing dangerous handling of boiling water until the procedural sequence is firmly consolidated.
9. Measurement & Assessment
Accurate measurement during backward chaining protocols is essential for making data-based instructional decisions. Practitioners employ specialized behavioral measurement systems structured around the task analysis:
Assessment begins with a comprehensive baseline evaluation, typically executed via either a single-opportunity probe or a multiple-opportunity probe method. In the single-opportunity probe, the learner is presented with the initial discriminative stimulus and given the opportunity to perform the chain. The moment an error is made, a step is omitted, or a predetermined latency threshold elapses, the assessment trial is terminated immediately, and all subsequent steps are scored as incorrect. This method minimizes the risk of inadvertently teaching errors during assessment, though it tends to underestimate the learner’s true repertoire if they possess mastery of later isolated steps.
In the multiple-opportunity probe, if the learner fails to execute a step correctly within a specified latency, the practitioner performs that step out of view or prompts it minimally without reinforcement, setting up the environmental conditions for the subsequent step. The learner is then scored on their ability to execute the next component. This approach provides a granular baseline of every step across the task analysis, identifying pre-existing splinter skills that can be leveraged using backward chaining with leaps ahead.
During ongoing intervention, data collection monitors the level of independence on each step using a standardized prompt hierarchy scale (e.g., 0 = Independent, 1 = Gestural Prompt, 2 = Verbal Prompt, 3 = Model Prompt, 4 = Physical Prompt). These scores are plotted on cumulative acquisition charts or standard time-series line graphs. Practitioners inspect the trend, level, and variability of unprompted correct responses specifically for the training link currently targeted by the backward progression.
10. Applications & Practical Significance
Backward chaining demonstrates extensive clinical, educational, vocational, and technological utility. In pediatric populations with developmental and intellectual disabilities, it represents the gold standard for establishing functional adaptive living skills (activities of daily living, or ADLs). Skills such as handwashing, dental hygiene, toileting sequences, and garment donning are universally taught via backward chaining because the natural conclusion of each routine provides an immediate sensory shift (e.g., clean hands, clear mouth, warm body) that functions as a natural reinforcer.
In vocational rehabilitation, backward chaining enables adolescents and adults with cognitive challenges to acquire complex assembly, packaging, culinary, and clerical repertoires. For instance, in an industrial setting requiring the collation, folding, inserting, and sealing of mailing packages, an employee taught via backward chaining first masters peeling the adhesive strip and sealing the envelope, progressing backward to paper collating. This ensures that every instructional episode yields a completed, commercially viable product, fostering vocational competence and reducing material waste.
In physical therapy and athletic coaching, complex kinematic movements are frequently shaped using backward chaining principles. A gymnastics coach teaching a complex vault or dismount may instruct an athlete to master the landing and stabilization phase first before systematically chaining the preceding aerial rotation, hurdle, and approach run. This mitigates injury risk and guarantees that the athlete possesses the necessary motor competencies to safely conclude the maneuver before executing the higher-velocity entry links.
Beyond behavioral domains, backward chaining serves as a cornerstone of formal inference engines in computer science and artificial intelligence. Rule-based expert systems—such as medical diagnostic systems, fault-detection algorithms, and legal reasoning software—employ backward chaining inference to evaluate hypotheses. Starting from a specified goal state (e.g., “Patient has Condition X”), the inference engine searches the knowledge base backward to identify whether the antecedent clinical evidence and diagnostic criteria support the conclusion.
11. Research & Empirical Evidence
Empirical literature spanning several decades substantiates the efficacy of backward chaining across diverse populations and behavioral typologies. Seminal comparative investigations have systematically evaluated backward chaining against alternative approaches such as forward chaining and total-task presentation.
Early research by Spooner and Spooner (1984) provided rigorous comparative analyses of chaining variants across individuals with severe intellectual disabilities. Their findings indicated that while total-task presentation is frequently efficient for learners with mild-to-moderate impairments who possess robust imitative repertoires, backward chaining consistently demonstrated superiority for complex multi-step motor tasks among individuals with profound intellectual impairments and severe attentional deficits. The authors attributed this advantage to the rapid, invariant contact with terminal reinforcement characteristic of backward protocols.
In a landmark empirical review, task analysis and chaining paradigms were evaluated by Browder and colleagues (2007) across literacy and functional life skill interventions. The synthesis established that backward chaining protocols yielded significantly higher effect sizes when teaching personal care and hygiene repertoires compared to non-systematic instructional strategies. The data indicated that backward progression minimized behavioral regressions and promoted superior skill maintenance at three- and six-month follow-ups, because the terminal link had undergone significantly more overlearning and reinforcement trials than any other link in the sequence.
Further experimental work in neurorehabilitation by Giles and Clark-Wilson (1993) demonstrated the clinical utility of backward chaining in patients with severe post-traumatic amnesia and frontal lobe damage. Their empirical trials revealed that patients who failed repeatedly under conventional verbal instruction successfully acquired multi-step instrumental activities of daily living (IADLs) when guided through backward chaining. The researchers noted that backward chaining capitalizes on preserved implicit and procedural memory systems, bypassing damaged explicit memory recall pathways.
12. Cultural & Cross-Cultural Considerations
The cross-cultural implementation of backward chaining requires clinicians and educators to account for cultural variations in behavioral topographies, developmental expectations, and the social validity of specific daily living routines. What constitutes an “essential” adaptive behavioral chain varies profoundly across cultural, socioeconomic, and geographic settings.
In many Western instructional environments, developmental paradigms prioritize rapid personal autonomy in individualistic self-help tasks, such as feeding oneself with specific cutlery or dressing in individualized apparel. However, in collectivistic cultural environments, familial caregiving practices may normative-developmentally extend caregiver-assisted feeding, bathing, and dressing well into middle childhood without regarding these dependencies as deficits. Imposing a rigid task analysis and backward chaining protocol on behaviors that a family considers culturally inappropriate or non-normative undermines intervention adherence and ecological validity.
Furthermore, the physical topography of target chains must be tailored culturally. A task analysis for food preparation or meal consumption must reflect culturally authentic utensils, dining postures, and food handling practices (e.g., eating with the right hand using flatbread versus utilizing forks and knives or chopsticks). Behavior analysts practicing cross-culturally must collaborate with families to co-construct the task analysis, ensuring that the terminal reinforcement and discriminative stimuli embedded within the chain resonate with the family’s cultural, linguistic, and domestic practices.
13. Criticisms, Debates & Limitations
Despite its documented empirical efficacy, backward chaining has been subject to theoretical debates, procedural criticisms, and clinical limitations within the behavioral and educational literature.
A primary criticism centers on instructional efficiency and time constraints. Because backward chaining requires the instructor to actively perform or physically prompt all non-targeted initial steps during every single instructional trial, it can be resource-intensive and physically demanding for practitioners. In classroom environments with high student-to-teacher ratios, conducting intensive one-on-one backward chaining sessions is frequently logistically challenging. Consequently, many school-based practitioners default to total-task presentation, where the learner attempts every step of the chain on every trial with varying levels of assistance, even if total-task presentation produces higher error rates for certain individuals.
A related instructional dilemma involves the potential suppression of emergent learning and learner agency. Critics argue that when instructors strictly complete all initial steps for the individual, they prevent the learner from demonstrating spontaneous, incidental acquisition of upstream skills. Backward chaining can inadvertently engender prompt dependency if the instructor fails to routinely probe the earlier links in the chain. The adaptation known as “backward chaining with leaps ahead” was developed precisely to mitigate this rigid constraint, yet it remains underutilized in clinical practice.
Finally, theoretical debates persist regarding which chaining paradigm produces the fastest rate of acquisition. While backward chaining is widely celebrated for its reinforcement proximity, empirical comparisons (e.g., Slocum & Tiger, 2011) have demonstrated that the relative efficacy of forward versus backward chaining is highly idiosyncratic across individual learners. Some participants acquire skills more rapidly via forward chaining because the initial step aligns naturally with the chronological flow of environmental antecedents, whereas others thrive exclusively under the reinforcement dynamics of backward chaining. As a consequence, rigid adherence to backward chaining without individualized comparative empirical assessment is viewed by contemporary behavior analysts as a clinical oversimplification.
14. Related Terms & Distinctions
Understanding backward chaining requires differentiating it clearly from related behavioral and computational constructs:
- Forward Chaining: An instructional protocol in which behaviors are taught in their natural chronological order, beginning with Step 1. The learner independently executes Step 1 to receive reinforcement, while subsequent unlearned steps are completed or prompted by the trainer; once Step 1 is mastered, instruction advances to Step 2. Unlike backward chaining, the learner does not contact the terminal natural reinforcer during early instructional stages.
- Total-Task Presentation (Concurrent Chaining): A variation of forward chaining where the learner is provided the opportunity to perform every step in the task analysis on every single trial, receiving assistance across all unmastered links. This is distinguished from backward chaining by its simultaneous exposure to all steps rather than isolating a single target link for reverse progression.
- Shaping: The differential reinforcement of successive approximations toward a single terminal target behavior. While chaining links distinct, separate behaviors into a functional sequence, shaping systematically alters the topography, duration, or intensity of a single discrete response.
- Task Analysis: The prerequisite process of breaking a complex skill down into smaller, individual, teachable behavioral units. Task analysis is not an instructional method itself, but rather the diagnostic and structural blueprint upon which backward, forward, or total-task chaining protocols are engineered.
- Forward Chaining (Artificial Intelligence): A data-driven inference approach in computer science that begins with known facts and applies conditional inference rules (IF-THEN) to extract new data and move forward toward a conclusion. This is the exact computational opposite of algorithmic backward chaining, which starts with a hypothesized goal and works backward to locate supporting facts.
15. Summary / Key Takeaways
Backward chaining represents one of the most methodologically rigorous and empirically supported instructional procedures within applied behavior analysis and neurorehabilitation. By deconstructing complex skills through meticulous task analysis and teaching the final step first, the procedure ensures that learners immediately contact the most potent, natural terminal reinforcement available. This immediate reinforcement proximity dramatically minimizes instructional frustration, reduces behavioral errors, and establishes each step in the chain as an alternating conditioned reinforcer and discriminative stimulus.
While backward chaining requires meticulous preparation, extensive one-on-one instructional investment, and cultural customization to avoid instructional mismatches, its effectiveness in establishing self-help, vocational, and motor repertoires in learners with severe developmental disabilities or neurological impairments remains unmatched. When implemented with systematic data collection, appropriate prompt fading hierarchies, and dynamic adaptations like leaping ahead, backward chaining provides an indispensable bridge from procedural dependence to autonomous behavioral mastery.
References
- Ayllon, T., & Azrin, N. (1968). The token economy: A motivational system for therapy and rehabilitation. Appleton-Century-Crofts.
- Browder, D. M., Trela, K., & Jimenez, B. (2007). Training teachers to follow a task analysis to engage students with moderate-to-severe intellectual disabilities in inclusive general education classrooms. Education and Training in Developmental Disabilities, 42(3), 294–308.
- Cooper, J. O., Heron, T. E., & Heward, W. L. (2020). Applied behavior analysis (3rd ed.). Pearson.
- Giles, G. M., & Clark-Wilson, J. (1993). Brain injury rehabilitation: A neurofunctional approach. Chapman & Hall.
- Gold, M. W. (1976). Task analysis of a complex assembly task by the retarded blind. Exceptional Children, 43(2), 78–84.
- Skinner, B. F. (1938). The behavior of organisms: An experimental analysis. Appleton-Century-Crofts.
- Slocum, S. K., & Tiger, J. H. (2011). An assessment of the relative efficiency of backward and forward chaining for teaching multi-step tasks to children with autism. Journal of Applied Behavior Analysis, 44(4), 793–805.
- Spooner, F., & Spooner, D. (1984). A review of chaining techniques: Implications for future research and practice. Education and Training of the Mentally Retarded, 19(2), 114–124.
- Wolery, M., Ault, M. J., & Doyle, P. M. (1992). Teaching students with moderate to severe disabilities: Use of response prompting strategies. Longman.