Using Process Data to Improve Classification Accuracy of Cognitive Diagnosis Model.

With the advance of computer-based assessments, many process data, such as response times (RTs), action sequences, Eye-tracking data, the log data for collaborative problem-solving (CPS) and mouse click/drag becomes readily available. Findings from previous studies (e.g., Peng et al., Multivariate Behavioral Research, 1-20, 2021; Xu, The British Journal of Mathematical and Statistical Psychology, 73(3), 474-505, 2020; He & von Davier, Handbook of research on technology tools for real-world skill development (pp. 750-777). IGI Global, 2016; Man & Harring, Educational and Psychological Measurement, 81(3), 441-465, 2021) suggest a substantial relationship between this human-computer interactive process information and proficiency, which means these process data were potentially useful variables for psychological and educational measurement. To make full use of the process data, this paper aims to combine two useful and easily available types of process data, including the mouse click/drag traces and the response times, to the conventional cognitive diagnostic model (CDM) to better understand individual's response behavior and improve the classification accuracy of existing CDM. Then the full Bayesian analysis using Markov chain Monte Carlo (MCMC) was employed to estimate the proposed model parameters. The viability of the proposed model was investigated by an empirical data and two simulation studies. Results indicated the proposed model combing both types of process data could not only improve the attribute classification reliability in real data analysis, but also provide an improvement on item parameters recovery and person classification accuracy.

Equivalence class analysis of responding consistent with the relational frame of opposition.

We replicated and extended studies showing that contextual cues for matching stimuli from 2 separate equivalence classes control the same derived relations as contextual cues for opposition frames in RFT studies. We conducted 2 experiments with 6 college students. In Phase 1, they received training in a conditional discrimination AB. Then, they received training for maintaining AB with X1 as context, and for reversing the sample-comparison relations of AB, with X2. In Phase 2, X1 functioned as context for matching same-class stimuli, and X2 functioned as context for matching separate-class stimuli. In Phase 3, X2 controlled the same derived arbitrary relations as cues for opposition frames in RFT studies. This functional equivalence may suggest that X2 functioned as a cue for opposition frames. In Phase 4, participants matched different stimuli with X2 as context, instead of matching most different (opposite) stimuli. In addition, Different, a cue for matching different stimuli, controlled the same derived arbitrary relations as X2. These results are incompatible with X2 being a cue for opposition frames. Contextual control over equivalence and responding by exclusion can explain these outcomes. The implications of these findings for RFT studies on opposition frames are discussed.

Differential derived stimulus relations across probe-trial versus adduction testing are not a function of comparison-stimulus presentation.

The present research assessed whether unreliable derivation of three-node equivalence relations in adduction testing is due to the absence of comparison-stimulus presentation. College students first learned arbitrary-matching-to-sample discriminations (AB, BC, CD, and DE) with Arabic numbers (A), nonsense syllables (B, C, D), and nonrepresentational forms (E) as stimuli. Participants then were exposed to either probe-trial testing or adduction testing. In probe-trial testing, each trial involved one of the four E stimuli (e.g., E1) as a sample and each of the four A stimuli (A1, A2, A3, A4) as comparisons. Successful performance involved participants selecting the corresponding A stimulus in the absence of differential consequences (e.g., A1 in the presence of E1). In adduction testing, each trial involved an arithmetic equation such that successful performance involved combining derived EA relations with simple math skills. Critically, unlike in typical adduction testing, participants were presented with four response options to select from (i.e., comparison stimuli). In probe-trial testing, each participant derived the EA relations, whereas no participant performed successfully in adduction testing. The present findings suggest that the unreliable derivation of three-node equivalence relations in adduction testing is not due to the typical absence of comparison stimuli. The present research challenges experimental psychologists to clarify the role of testing context in identifying the necessary and sufficient conditions that control emergent learning.

Temporal framing and the hidden-zero effect: rate-dependent outcomes on delay discounting.

Recent research suggests that presenting time intervals as units (e.g., days) or as specific dates, can modulate the degree to which humans discount delayed outcomes. Another framing effect involves explicitly stating that choosing a smaller-sooner reward is mutually exclusive to receiving a larger-later reward, thus presenting choices as an extended sequence. In Experiment 1, participants (N = 201) recruited from Amazon Mechanical Turk completed the Monetary Choice Questionnaire in a 2 (delay framing) by 2 (zero framing) design. Regression suggested a main effect of delay, but not zero, framing after accounting for other demographic variables and manipulations. We observed a rate-dependent effect for the date-framing group, such that those with initially steep discounting exhibited greater sensitivity to the manipulation than those with initially shallow discounting. Subsequent analyses suggest these effects cannot be explained by regression to the mean. Experiment 2 addressed the possibility that the null effect of zero framing was due to within-subject exposure to the hidden- and explicit-zero conditions. A new Amazon Mechanical Turk sample completed the Monetary Choice Questionnaire in either hidden- or explicit-zero formats. Analyses revealed a main effect of reward magnitude, but not zero framing, suggesting potential limitations to the generality of the hidden-zero effect.

Developing a novel preparation to analyze the onset of derived stimulus relations.

The experimental analysis of derived stimulus relations is a critical research area. A training-then-testing preparation nearly always is used to study derived relations. In the training phase, participants learn the relevant baseline discriminations via differential consequences (e.g., AB and AC relations). In the testing phase, they are presented with probe trials in the absence of differential consequences (e.g., BA and CA symmetry trials and BC and CB equivalence trials). High accuracy levels sometimes are observed from the start of testing such that it is unclear whether the participants learned these relations before testing. The present experiment reports data from a novel preparation that monitors the development of derived relations as trained relations are acquired. Three college students were presented with both training trials (AB, AC) and testing trials (BA, CA, BC, CB) in every session from the start of experimentation. Each participant learned each of the six discriminations by the end of experimentation. Most importantly, they learned the trained and symmetrical relations in close temporal proximity and the equivalence relations only after learning the symmetrical relations. These results are consistent with several findings demonstrating disparities between learning different forms of derived relational responding. The results validate the utility of the present preparation in the experimental analysis of derived relational learning.

Attention and Information Acquisition: Comparison of Mouse-Click with Eye-Movement Attention Tracking.

Attention is crucial as a fundamental prerequisite for perception. The measurement of attention in viewing and recognizing the images that surround us constitutes an important part of eye movement research, particularly in advertising-effectiveness research. Recording eye and gaze (i.e. eye and head) movements is considered the standard procedure for measuring attention. However, alternative measurement methods have been developed in recent years, one of which is mouse-click attention tracking (mcAT) by means of an on-line based procedure that measures gaze motion via a mouse-click (i.e. a hand and finger positioning maneuver) on a computer screen. Here we compared the validity of mcAT with eye movement attention tracking (emAT). We recorded data in a between subject design via emAT and mcAT and analyzed and compared 20 subjects for correlations. The test stimuli consisted of 64 images that were assigned to eight categories. Our main results demonstrated a highly significant correlation (p < 0.001) between mcAT and emAT data. We also found significant differences in correlations between different image categories. For simply structured pictures of humans or animals in particular, mcAT provided highly valid and more consistent results compared to emAT. We concluded that mcAT is a suitable method for measuring the attention we give to the images that surround us, such as photographs, graphics, art or digital and print advertisements.

Transfer of function and prior derived-relations testing.

This experiment assessed transfer of function through equivalence relations with and without prior derived-stimulus-relations (DSR) testing. In a DSR-Testing Group, eight college students learned A-B and A-C discriminations in baseline. They then derived the B-C and C-B equivalence relations before being exposed to a transfer-of-function manipulation and test. Eight participants in a No-DSR Testing Group were exposed to the transfer-of-function manipulation and test immediately after learning the baseline discriminations (i.e., B-C and C-B testing were omitted). In the transfer-of-function manipulation, participants learned to respond differently in the presence of B1 and B2 to avoid money loss. In the transfer-of-function test, responding in the presence of C1 and C2 was measured in the absence of differential consequences. Transfer of function occurred reliably only in the DSR-Testing Group (i.e., participants responding to C1 and C2 in the manner they learned to respond to B1 and B2, respectively). These findings support the notion that prior DSR testing can be critical to observing transfer of function.

Adduction of untested derived stimulus relations depends on environmental complexity.

The present research assessed adduction involving derived stimulus relations as a function of environmental complexity. In Group CA, four college students were trained with arbitrary-matching-to-sample discriminations that could have established four, 3-member stimulus classes. In Group EA, four other students were trained with discriminations that could have established four, 5-member classes. Neither group received derived-relations testing; instead, adduction was assessed immediately after the baseline discriminations were learned. The adduction assessment required participants to derive the untested CA (Group CA) or EA (Group EA) equivalence relations and combine them with their already learned math skills. All participants in Group CA showed above 90% accuracy during the adduction assessment, whereas only one of four Group EA participants responded in that manner. These results extend adduction to untested equivalence relations and clarify the environmental conditions under which such adduction is less likely to occur (i.e., with larger relational networks).

Understanding behavior under nonverbal transitive-inference procedures: Stimulus-control-topography analyses.

Following training with verbal stimulus relations involving A is greater than B and B is greater than C, verbally-competent individuals reliably select A>C when asked "which is greater, A or C?" (i.e., verbal transitive inference). This result is easy to interpret. Nonhuman animals and humans with and without intellectual disabilities have been exposed to nonverbal transitive-inference procedures involving trained arbitrary stimulus relations. Following the training of A+B-, B+C-, C+D-, and D+E-, B reliably is selected over D (i.e., nonverbal transitive inference). Such findings are more challenging to interpret. The present research explored accounts of nonverbal transitive inference based in transitive inference per se, reinforcement, such as value-transfer theory, and operant stimulus control. In Experiment 1, college students selected B>G following the training of A+B-, B+C-, C+D-///E+F-, F+G-, and G+H- (where///signifies the omission of D+E-). In Experiment 2, college students selected B>G following the training of A+B-, B+C-, C+D-///E+F-, F+G-, and G+X- (where X refers to 10 stimuli that alternated across trials). In Experiment 3, college students selected G>B following the training of Y+B-, B+C-, C+D-///E+F-, F+G-, and G+X- (where Y and X refer to 10 stimuli, respectively, that alternated across trials). These findings are discussed in the context of operant stimulus control by offering an approach based in stimulus B typically acquiring only a select stimulus control topography.

Effects of differential rates of alternative reinforcement on resurgence of human behavior.

Despite the success of exposure-based psychotherapies in anxiety treatment, relapse remains problematic. Resurgence, the return of previously eliminated behavior following the elimination of an alternative source of reinforcement, is a promising model of operant relapse. Nonhuman resurgence research has shown that higher rates of alternative reinforcement result in faster, more comprehensive suppression of target behavior, but also in greater resurgence when alternative reinforcement is eliminated. This study investigated rich and lean rates of alternative reinforcement on response suppression and resurgence in typically developing humans. In Phase 1, three groups (Rich, n = 18; Lean, n = 18; Control, n = 10) acquired the target response. In Phase 2, target responding was extinguished and alternative reinforcement delivered on RI 1 s, RI 3 s, and extinction schedules, respectively. Resurgence was assessed during Phase 3 under extinction conditions for all groups. Target responding was suppressed most thoroughly in Rich and partially in Lean. Target responding resurged in the Rich and Lean groups, but not in the Control group. Between groups, resurgence was more pronounced in the Rich group than the Lean and Control groups. Clinical implications of these findings, including care on the part of clinicians when identifying alternative sources of reinforcement, are discussed.

Flexible Piezoelectric Energy Harvesting from Mouse Click Motions.

In this paper, we study energy harvesting from the mouse click motions of a robot finger and a human index finger using a piezoelectric material. The feasibility of energy harvesting from mouse click motions is experimentally and theoretically assessed. The fingers wear a glove with a pocket for including the piezoelectric material. We model the energy harvesting system through the inverse kinematic framework of parallel joints in a finger and the electromechanical coupling equations of the piezoelectric material. The model is validated through energy harvesting experiments in the robot and human fingers with the systematically varying load resistance. We find that energy harvesting is maximized at the matched load resistance to the impedance of the piezoelectric material, and the harvested energy level is tens of nJ.

Resurgence of target responding does not exceed increases in inactive responding in a forced-choice alternative reinforcement procedure in humans.

Resurgence following removal of alternative reinforcement has been studied in non-human animals, children with developmental disabilities, and typically functioning adults. Adult human laboratory studies have included responses without a controlled history of reinforcement, included only two response options, or involved extensive training. Arbitrary responses allow for control over history of reinforcement. Including an inactive response never associated with reinforcement allows the conclusion that resurgence exceeds extinction-induced variability. Although procedures with extensive training produce reliable resurgence, a brief procedure with the same experimental control would allow more efficient examination of resurgence in adult humans. We tested the acceptability of a brief, single-session, three-alternative forced-choice procedure as a model of resurgence in undergraduates. Selecting a shape was the target response (reinforced in Phase I), selecting another shape was the alternative response (reinforced in Phase II), and selecting a third shape was never reinforced. Despite manipulating number of trials and probability of reinforcement, resurgence of the target response did not consistently exceed increases in the inactive response. Our findings reiterate the importance of an inactive control response and call for reexamination of resurgence studies using only two response options. We discuss potential approaches to generate an acceptable, brief human laboratory resurgence procedure.

Contextual control using a go/no-go procedure with compound abstract stimuli.

Contextual control has been described as (1) a five-term contingency, in which the contextual stimulus exerts conditional control over conditional discriminations, and (2) allowing one stimulus to be a member of different equivalence classes without merging them into one. Matching-to-sample is the most commonly employed procedure to produce and study contextual control. The present study evaluated whether the go/no-go procedure with compound stimuli produces equivalence classes that share stimuli. This procedure does not allow the identification of specific stimulus functions (e.g., contextual, conditional, or discriminative functions). If equivalence classes were established with this procedure, then only the latter part of the contextual control definition (2) would be met. Six undergraduate students participated in the present study. In the training phases, responses to AC, BD, and XY compounds with stimuli from the same classes were reinforced, and responses to AC, BD, and XY compounds with stimuli from different classes were not. In addition, responses to X1A1B1, X1A2B2, X2A1B2, and X2A2B1 compounds were reinforced and responses to the other combinations were not. During the tests, the participants had to respond to new combinations of stimuli compounds YCD to indicate the formation of four equivalence classes that share stimuli: X1A1B1Y1C1D1, X1A2B2Y1C2D2, X2A1B2Y2C1D2, and X2A2B1Y2C2D1. Four of the six participants showed the establishment of these classes. These results indicate that establishing contextual stimulus functions is unnecessary to produce equivalence classes that share stimuli. Therefore, these results are inconsistent with the first part of the definition of contextual control.

Failure to observe untested derived stimulus relations in extinction: implications for understanding stimulus-equivalence formation.

Wilson and Hayes (1996) and Doughty, Kastner, and Bismark (2011) observed resurgence of past equivalence relations when newer equivalence relations were punished or extinguished, respectively. Their findings support the notion that deriving stimulus relations is a form of operant behavior. Although there is consensus regarding the operant nature of deriving untrained stimulus relations, the necessary and sufficient conditions that establish these relations remain unclear. For example, in the aforementioned work, the resurgent equivalence relations were tested earlier in each study. The present research investigated whether this resurgence of equivalence relations requires their initial testing. In Experiment 1, college students received arbitrary matching-to-sample training in Phase 1. After these baseline discriminations were established, equivalence testing was omitted. In Phase 2, four 4-member equivalence classes were established that were inconsistent with the Phase-1 training. These Phase-2 relations then were extinguished to test whether the equivalence relations consistent with Phase-1 training would occur. Unlike in earlier research, these untested relations did not occur reliably. This finding was replicated in Experiment 2, and additional tests of stimulus relatedness did not reveal any evidence of derived stimulus relations consistent with the Phase-1 training. The present research extends other findings suggesting that the equivalence-testing context helps establish these stimulus classes.

Merging separately established stimulus classes with outcome-specific reinforcement.

This study extended previous research on equivalence relations established with outcome-specific reinforcers to include the merger of separately established stimulus classes. Participants were four adults. Conditional discriminations AC and BC were trained first. Correct selections of C1 (C2, or C3) in the presence of A1 or B1 (A2 or B2, or A3 or B3) were followed by red (blue, or white) tokens; tokens were exchanged for value added to three participant-selected gift cards. Outcomes on equivalence tests for three-member classes ABC were positive. DF and EF were trained with the same reinforcing consequences, and tests were positive for three-member classes DEF. Results of class merger tests with combinations of stimuli from the ABC and DEF classes (AD, FB, etc.) were immediately positive for two participants, demonstrating six-member classes ABCDEF with reinforcers as nodes. Merger tests for a third participant were initially negative but became positive after brief exposure to unreinforced probe trials with reinforcers as comparison stimuli. Following class merger, tests for matching the reinforcers to samples and comparisons were also positive. Class-merger test results were negative for a fourth participant. The results provide the first demonstration of eight-member equivalence classes including two outcome-specific conditioned reinforcing stimuli.

Facilitating relational framing in children and individuals with developmental delay using the relational completion procedure.

The Relational Completion Procedure is effective for establishing same, opposite and comparative derived relations in verbally able adults, but to date it has not been used to establish relational frames in young children or those with developmental delay. In Experiment 1, the Relational Completion Procedure was used with the goal of establishing two 3-member sameness networks in nine individuals with Autism Spectrum Disorder (eight with language delay). A multiple exemplar intervention was employed to facilitate derived relational responding when required. Seven of nine participants in Experiment 1 passed tests for derived relations. In Experiment 2, eight participants (all of whom, except one, had a verbal repertoire) were given training with the aim of establishing two 4-member sameness networks. Three of these participants were typically developing young children aged between 5 and 6 years old, all of whom demonstrated derived relations, as did four of the five participants with developmental delay. These data demonstrate that it is possible to reliably establish derived relations in young children and those with developmental delay using an automated procedure.

Optimal and nonoptimal choice in a laboratory-based sunk cost task with humans: a cross-species replication.

The current four experiments examined the sunk cost effect-nonoptimal persistence following investment-in a laboratory-based decision-making task with adult humans. Subjects made repeated decisions about whether to persist in a course of action-a fixed-ratio schedule whose response requirements varied unpredictably from one trial to the next-or to abandon it and escape in favor of a new trial with a potentially smaller fixed ratio schedule. Satisfying the ratio requirement produced a brief video clip from a preferred television program. In Experiment 1, sunk-cost errors were less likely in subjects who had previously experienced markedly differential reinforcement for escape. In Experiment 2, stimulus changes correlated with changes in mean response requirement, and these changes reduced sunk-cost errors in a small number of subjects. In Experiment 3, sunk-cost errors occurred more frequently as the ratio of the mean response requirements for persistence and escape approached 1.0. The importance of this variable was further supported by the results of Experiment 4, in which features other than this ratio did not markedly alter performance. These four experiments identified some key determinants of whether humans commit the sunk-cost error and confirmed the utility of video clips as reinforcers in experimental research with humans.

Evaluating merger and intersection of equivalence classes with one member in common.

Sidman (1994) noted that the existence of a member that is common to more than one class may produce either class merger (union) or class intersection. A multiple-selection, matching-to-sample test was developed to examine the conditions under which these outcomes occur. Test trials each required three conditional discriminations involving selection or rejection of comparison stimuli under control of samples representing two categories. Test results obtained from an initial group of typical adults using familiar stimuli (DOG and BIRD, pictures of dogs and birds and relevant printed breed names (e.g., DALMATIAN, RETRIEVER) showed the conditional stimulus control best described as intersection. For example, the word DALMATIAN provided the context for selecting the dalmatian but not the retriever picture. However, these results may have depended on the participants' verbal history as English speakers. Would conditional-discrimination training with overlapping sets of laboratory-generated stimuli also result in intersection? Naïve typical adults were assigned to one of three different training conditions. Like the participants tested with familiar stimuli, these participants demonstrated highly reliable test outcomes best described as showing class intersection, regardless of training condition. These findings begin to elucidate the necessary and sufficient conditions for establishing complex category-like classes of stimuli.

Operant measurement of auditory threshold in prelingually deaf users of cochlear implants: II

Two experiments evaluated an operant procedure for establishing stimulus control using auditory and electrical stimuli as a baseline for measuring the electrical current threshold of electrodes implanted in the cochlea. Twenty-one prelingually deaf children, users of cochlear implants, learned a Go/No Go auditory discrimination task (i.e., pressing a button in the presence of the stimulus but not in its absence). When the simple discrimination baseline became stable, the electrical current was manipulated in descending and ascending series according to an adapted staircase method. Thresholds were determined for three electrodes, one in each location in the cochlea (basal, medial, and apical). Stimulus control was maintained within a certain range of decreasing electrical current but was eventually disrupted. Increasing the current recovered stimulus control, thus allowing the determination of a range of electrical currents that could be defined as the threshold. The present study demonstrated the feasibility of the operant procedure combined with a psychophysical method for threshold assessment, thus contributing to the routine fitting and maintenance of cochlear implants within the limitations of a hospital setting.(AU)

Operant measurement of auditory threshold in prelingually deaf users of cochlear implants: II

Two experiments evaluated an operant procedure for establishing stimulus control using auditory and electrical stimuli as a baseline for measuring the electrical current threshold of electrodes implanted in the cochlea. Twenty-one prelingually deaf children, users of cochlear implants, learned a Go/No Go auditory discrimination task (i.e., pressing a button in the presence of the stimulus but not in its absence). When the simple discrimination baseline became stable, the electrical current was manipulated in descending and ascending series according to an adapted staircase method. Thresholds were determined for three electrodes, one in each location in the cochlea (basal, medial, and apical). Stimulus control was maintained within a certain range of decreasing electrical current but was eventually disrupted. Increasing the current recovered stimulus control, thus allowing the determination of a range of electrical currents that could be defined as the threshold. The present study demonstrated the feasibility of the operant procedure combined with a psychophysical method for threshold assessment, thus contributing to the routine fitting and maintenance of cochlear implants within the limitations of a hospital setting.