Parameter Estimation By Sequential Testing Research Articles

The perception of synthetic speech versus natural speech stimuli in adult listeners

The majority of speech perception research investigates interpersonal perception through presentation of two types of stimuli: synthetic speech and/or natural speech. Synthetically created speech stimuli allows for precisely controlled manipulation of individual acoustic variables, but can sound unnatural calling into questions its ecological validity. Alternatively, natural speech preserves the full complement of redundant acoustic cues yet limits control of specific acoustic parameters. In this study, we investigate speech discrimination comparing perception of specified acoustic parameters (formant transition duration, F3 onset frequency, F3-F2 distance) across three types of speech stimuli: synthetic speech stimuli, natural speech of a standard adult speaker, and each participant’s own voice. We use the Wide Range Acoustic Accuracy Scale (WRAAS), a computer-based perceptual assessment that uses a parameter estimation by sequential testing (PEST) adaptive-tracking algorithm to rapidly identify a listener’s ability to discriminate sounds. Most trials are presented near the Just-Noticeable-Difference (JND) of a stimulus continuum; thus, few trials are “wasted” with obviously same or different sounds. This study of typical adult listeners (n = 20, enrollment is ongoing) will provide baseline data for future work investigating perception of these same stimuli in children.

Quantifying the streamflow response to groundwater abstractions for irrigation or drinking water at catchment scale using SWAT and SWAT\u2013MODFLOW

BackgroundGroundwater abstraction can cause a decline in the water table, and thereby affects surface streamflow connected to the aquifer, which may impair the sustainability of both the water resource itself and the ecosystem that it supports. To quantify the streamflow response to groundwater abstractions for either irrigation or drinking water at catchment scale and compared the performance of the widely used semi-distributed hydrological model SWAT and an recently integrated surface–subsurface model SWAT–MODFLOW, we applied both SWAT and SWAT–MODFLOW to a groundwater-dominated catchment in Denmark and tested a range of groundwater abstraction scenarios.ResultsTo accommodate the study area characteristics, the SWAT–MODFLOW model complex was further developed to enable the Drain package and an auto-irrigation routine to be used. A PEST (parameter estimation by sequential testing)-based approach which enables simultaneous calibration of SWAT and MODFLOW parameters was developed to calibrate SWAT–MODFLOW. Both models demonstrated generally good statistical performance for the temporal pattern of streamflow, with better R2 and NSE (Nash–Sutcliffe efficiency) for SWAT–MODFLOW but slightly better PBIAS (percent bias) for SWAT. Both models indicated that drinking water abstractions caused some degree of streamflow depletion, while abstractions for returned irrigation led to a slight total flow increase, but may influence the hydrology outside the catchment. However, the streamflow decrease caused by drinking water abstractions simulated by SWAT was unrealistically low, and the streamflow increase caused by irrigation abstractions was exaggerated compared with SWAT–MODFLOW.ConclusionWe conclude that the SWAT–MODFLOW model produces much more realistic signals relative to the SWAT model when quantifying the streamflow response to groundwater abstractions for irrigation or drinking water; hence, it has great potential to be a useful tool in the management of water resources in groundwater-dominated catchments. With further development of SWAT–MODFLOW and the PEST-based approach developed for its calibration, this study would broaden the SWAT–MODFLOW application and benefit catchment managers.

Deep Brain Stimulation of the Pedunculopontine Tegmental Nucleus (PPN) Influences Visual Contrast Sensitivity in Human Observers.

The parapontine nucleus of the thalamus (PPN) is a neuromodulatory midbrain structure with widespread connectivity to cortical and subcortical motor structures, as well as the spinal cord. The PPN also projects to the thalamus, including visual relay nuclei like the LGN and the pulvinar. Moreover, there is intense connectivity with sensory structures of the tegmentum in particular with the superior colliculus (SC). Given the existence and abundance of projections to visual sensory structures, it is likely that activity in the PPN has some modulatory influence on visual sensory selection. Here we address this possibility by measuring the visual discrimination performance (luminance contrast thresholds) in a group of patients with Parkinson’s Disease (PD) treated with deep-brain stimulation (DBS) of the PPN to control gait and postural motor deficits. In each patient we measured the luminance-contrast threshold of being able to discriminate an orientation-target (Gabor-grating) as a function of stimulation frequency (high 60Hz, low 8/10, no stimulation). Thresholds were determined using a standard staircase-protocol that is based on parameter estimation by sequential testing (PEST). We observed that under low frequency stimulation thresholds increased relative to no and high frequency stimulation in five out of six patients, suggesting that DBS of the PPN has a frequency-dependent impact on visual selection processes at a rather elementary perceptual level.

Directional acuity of whole-body perturbations during standing balance

The ability to perceive the direction of whole-body motion during standing may be critical to maintaining balance and preventing a fall. Our first goal was to quantify kinesthetic perception of whole-body motion by estimating directional acuity thresholds of support-surface perturbations during standing. The directional acuity threshold to lateral deviations in backward support-surface motion in healthy, young adults was quantified as 9.5±2.4° using the psychometric method (n=25 subjects). However, inherent limitations in the psychometric method, such as a large number of required trials and the predetermined stimulus set, may preclude wider use of this method in clinical populations. Our second goal was to validate an adaptive algorithm known as parameter estimation by sequential testing (PEST) as an alternative threshold estimation technique to minimize the required trial count without predetermined knowledge of the relevant stimulus space. The directional acuity threshold was estimated at 11.7±3.8° from the PEST method (n=11 of 25 subjects, psychometric threshold=10.1±3.1°) using only one-third the number of trials compared to the psychometric method. Furthermore, PEST estimates of the direction acuity threshold were highly correlated with the psychometric estimates across subjects (r=0.93) suggesting that both methods provide comparable estimates of the perceptual threshold. Computational modeling of both techniques revealed similar variance in the estimated thresholds across simulations of about 1°. Our results suggest that the PEST algorithm can be used to more quickly quantify whole-body directional acuity during standing in individuals with balance impairments.

Development and validation of a novel tool for assessing pitch discrimination.

To develop and validate a tool for pitch discrimination that can be easily applied in otolaryngology clinics. Twenty healthy subjects performed three psychoacoustic tasks for assessing pitch discrimination thresholds, which were compared among the three tasks. In the pitch direction discrimination (PDirD) task, the subjects judged the pitch of two tones as higher or lower. In the pitch difference discrimination (PDiffD) task, the subjects judged the pitch of two tones as same or different. In the pitch pattern detection (PPD) task, the subjects judged the pattern of a tone complex. Using tone sets from level 0 to level 8, pitch thresholds were determined using the Parameter Estimation by Sequential Testing (PEST) staircase procedure. The tasks were performed without prior training. Every test was performed twice for each ear, and the better results were adopted as the threshold values. The PDirD thresholds varied considerably among the subjects and showed significantly larger deviations relative to those of the PPD thresholds. The number of below 0 thresholds, which are considered inadequate results, was significantly larger in the PDiffD task. In the PPD task, all thresholds except one were above level 6. The average total time taken to complete the PPD task was 7.7min, which was significantly lesser than that taken to complete the other two tasks. The results of our study suggest that the PPD task is a reliable and easily available tool for the assessment of pitch discrimination.

PSYCHOACOUSTICS: a comprehensive MATLAB toolbox for auditory testing.

PSYCHOACOUSTICS is a new MATLAB toolbox which implements three classic adaptive procedures for auditory threshold estimation. The first includes those of the Staircase family (method of limits, simple up-down and transformed up-down); the second is the Parameter Estimation by Sequential Testing (PEST); and the third is the Maximum Likelihood Procedure (MLP). The toolbox comes with more than twenty built-in experiments each provided with the recommended (default) parameters. However, if desired, these parameters can be modified through an intuitive and user friendly graphical interface and stored for future use (no programming skills are required). Finally, PSYCHOACOUSTICS is very flexible as it comes with several signal generators and can be easily extended for any experiment.

A comparison of relative-frequency and threshold-hunting methods to determine stimulus intensity in transcranial magnetic stimulation

ObjectiveStimulation intensity (SI) in transcranial magnetic stimulation is commonly set in relation to motor threshold (MT), or to achieve a motor-evoked potential (MEP) of predefined amplitude (usually 1mV). Recently, IFCN recommended adaptive threshold-hunting over the previously endorsed relative-frequency method. We compared the Rossini–Rothwell (R–R) relative-frequency method to an adaptive threshold-hunting method based on parameter estimation by sequential testing (PEST) for determining MT and the SI to target a MEP amplitude of 1mV (I1mV). MethodsIn 10 healthy controls we determined MT and I1mV with R–R and PEST using a blinded crossover design, and performed within-session serial PEST measurements of MT. ResultsThere was no significant difference between methods for MT (52.6±2.6% vs. 53.7±3.1%; p=0.302; % maximum stimulator output; R–R vs. PEST, respectively) or I1mV (66.7±3.0% vs. 68.8±3.8%; p=0.146). There was strong correlation between R–R and PEST estimates for both MT and I1mV. R–R required significantly more stimuli than PEST. Serial measurements of MT with PEST were reproducible. ConclusionsPEST has the advantage of speed without sacrificing precision when compared to the R–R method, and is adaptable to other SI targets. SignificanceOur results in healthy controls add to increasing evidence in favour of adaptive threshold-hunting methods for determining SI.

Experiments on Stochastic Resonance Toward Human Mimetic Tactile Data Processing

In the present research, human tactile stochastic resonance (SR) capable of enhancing sensitivity by superimposing proper noise upon undetectable weak signals is utilized to enhance the tactile processing method for social robotics. We develop an experimental apparatus composed of a piezoelectric actuator and its controller, and generate a step several microns high mixed with noise to perform a series of psychophysical experiments. Since psychophysical experiments are conducted based on the Parameter Estimation by Sequential Testing (PEST) method, we produce a PEST program that generates a stimuli sequence based on PEST. The experimental result shows that variation in the difference threshold (Difference Limen; DL) has a local minimum point in the relationship between DL and noise. Therefore, the tactile sensation’s just noticeable difference (JND) is decreased by appropriate external noise. Since JND denotes the scale divisions of sensation in the human mind, the present result shows that precise tactile sensations are enhanced by the appropriate external noise. Finally, we introduce a neural network model composed of nonlinear neurons with the bi-stable equilibrium condition to explain this result. Although original sensor data do not represent the morphology of the fine texture, the neural network model extracts the morphology and distinguishes the wave amplitude of the fine texture.

Measurement of contrast thresholds of chimpanzees using a Parameter Estimation by Sequential Testing (PEST) procedure(Summary of Awarded Presentations at the 24th Annual Meeting)

Measurement of contrast thresholds of chimpanzees using a Parameter Estimation by Sequential Testing (PEST) procedure(Summary of Awarded Presentations at the 24th Annual Meeting)

Estimating Resting Motor Thresholds in Transcranial Magnetic Stimulation Research and Practice

Resting motor threshold is the basic unit of dosing in transcranial magnetic stimulation (TMS) research and practice. There is little consensus on how best to estimate resting motor threshold with TMS, and only a few tools and resources are readily available to TMS researchers. The current study investigates the accuracy and efficiency of 5 different approaches to motor threshold assessment for TMS research and practice applications. Computer simulation models are used to test the efficiency and accuracy of 5 different adaptive parameter estimation by sequential testing (PEST) procedures. For each approach, data are presented with respect to the mean number of TMS trials necessary to reach the motor threshold estimate as well as the mean accuracy of the estimates. A simple nonparametric PEST procedure appears to provide the most accurate motor threshold estimates, but takes slightly longer (on average, 3.48 trials) to complete than a popular parametric alternative (maximum likelihood PEST). Recommendations are made for the best starting values for each of the approaches to maximize both efficiency and accuracy. In light of the computer simulation data provided in this article, the authors review and suggest which techniques might best fit different TMS research and clinical situations. Lastly, a free user-friendly software package is described and made available on the world wide web that allows users to run all of the motor threshold estimation procedures discussed in this article for clinical and research applications.

Discrimination of phoneme length differences in word and sentence contexts

The ability of listeners to discriminate phoneme duration differences within word and sentence contexts was measured. This investigation was part of a series of studies examining the audibility and perceptual importance of speech modifications produced by stuttering intervention techniques. Just noticeable differences (jnd’s) of phoneme lengths were measured via the parameter estimation by sequential testing (PEST) task, an adaptive tracking procedure. The target phonemes were digitally manipulated to vary from normal (130 m) to prolonged (210 m) duration in 2-m increments. In the first condition the phonemes were embedded in words. In the second condition the phonemes were embedded within words, which were further embedded in sentences. A four-interval forced-choice (4IAX) task was employed on each trial, and the PEST procedure determined the duration at which each listener correctly detected a difference between the normal duration and the test duration 71% of the time. The results revealed that listeners were able to reliably discriminate approximately 15-m differences in word context and 10-m differences in sentence context. An independent t-test showed a difference in discriminability between word and sentence contexts to be significant. These results indicate that duration differences were better perceived within a sentence context.

An Alternative Perspective on von Winterfeldt et al.'s (1997) Test of Consequence Monotonicity.

D. von Winterfeldt, N.-K. Chung, R. D. Luce, and Y. Cho (1997) provided several tests for consequence monotonicity of choice or judgment, using certainty equivalents of gambles. The authors reaxiomatized consequence monotonicity in a probabilistic framework and reanalyzed von Winterfeldt et al.'s main experiment via a bootstrap method. Their application offers new insights into consequence monotonicity as well as into von Winterfeldt et al.'s 3 experimental paradigms: judged certainty equivalents (JCE), QUICKINDIFF, and parameter estimation by sequential testing (PEST). For QUICKINDIFF, the authors found no indication of violations of "random consequence monotonicity." This sharply contrasts the findings of von Winterfeldt et al., who concluded that axiom violations were the most pronounced under that procedure. The authors found potential evidence for violations in JCE and certainty equivalents derived from PEST.

The remediation of clumsiness. I: An evaluation of Laszlo's kinaesthetic approach.

The effectiveness of a kinaesthetic training programme proposed by Laszlo for children with movement difficulties was evaluated by comparing two groups of 10 'clumsy' children matched pairwise on age, IQ and sex as well as degree of kinaesthetic and motor impairment. Tests of kinaesthetic ability, using the Parameter Estimation by Sequential Testing (PEST) procedure, and motor competence administered before and after treatment revealed an improvement in both groups on all measures but no differential effect of the training programme. Immediately after training, the changes we obtained in motor performance were confined to balance skills but, at follow up, 3 months later, changes in manual and ball skills were also evident. This unusual pattern of change requires replication. Our findings forced us to consider the possibility that any effect of Laszlo's recommended training programme had been obscured by our use of the PEST procedure, which had in itself facilitated motor learning.

Computer Methods for On-Line Hearing Testing with Auditory Brain Stem Responses

On-line computer methods were developed and evaluated for automated hearing testing with auditory brain stem responses (ABR). The system contained (1) a response recognition unit which recognized the presence or absence of an ABR response and (2) a threshold tracking unit which adjusted stimulus intensity and determined the ABR threshold. Two methods, one nontemplate (variance ratio) and one template (cross-correlation), were evaluated for response recognition while three threshold tracking methods were explored: clinical, Békésy, and PEST (parameter estimation by sequential testing). Both response recognition and threshold tracking units were evaluated by computer simulations for on-line operation using a standard set of experimentally recorded ABRs. The results indicated that while all the response recognition methods were reasonably accurate, their on-line use resulted in significant differences in test efficiency due to differences in tracking methods.

Exploring the development of area judgements using a PEST technique

Previous studies of the information integration rules used in area judgements have employed either rating scales, or area‐matching techniques in which a pair of rectangles is presented and one is adjusted to match the other in size. In this study a PEST (Parameter Estimation by Sequential Testing) procedure was used to track the point of subjective equality between a standard rectangle and a comparator rectangle of different shape. This method has the advantage that a simple binary choice is the only response required. Functional measurement analysis was carried out on groups of adults, 5‐ and 10‐year‐old children who each completed two replications of a 3 × 3 design in which 3 values of standard area were combined with 3 values of comparator width. The results supported previous findings that young children generally use an addition (height + width) rule whereas older children and adults use a multiplying rule in area judgement. A second experiment confirmed the difference in predominant strategy between 5‐ and 11‐year‐old children and eliminated one possible artefact due to the starting value of the tracking series. This confirmation of the development change in area‐judgement strategies using a different method suggests that the strategy changes are real and not an artefact of measurement techniques. In addition, the present technique reveals more widespread use of the multiplying strategy than previous methods based on rating scales. The implications of this finding are discussed.

The effect on the preference-reversal phenomenon of using choice indifferences

Preference reversal appears to be non-transitive choice behavior, inasmuch as subjects choosing between two gambles with similar expected values typically select the one with a larger chance of winning (a P-bet), yet they place a higher certainty equivalent, or judged indifference point (JIP), on the one with the larger amount to win (the $-bet). In the present experiments, we explore a pair of methods for determining money indifference points that are procedurally closer than is common in this research to the choice phase of the preference reversal experiment. The two new approaches for determining money indifference points for gambles (called choice indifference points, or CIP) are as follows: a version of the classical psychophysical up-down method and a variant called PEST (i.e., Parameter Estimation by Sequential Testing). With both methods, CIPs reduced the number of reversals, as compared to JIPs, to the point with PEST that the remaining ones may be due primarily to chance fluctuations in choices between gambles. Subjects who reversed frequently usually judged the gambles to have similar values; those who did not reverse at all, judged them as different. The typical subjects exhibits JIP>CIP for $-bets, but not for P-bets. This bias appears to underlie the observed intransitivity when JIP is used.

A comparison of three adaptive psychophysical procedures using inexperienced listeners.

Thresholds for a masked 1-kHz tone were obtained in single sessions from 60 inexperienced listeners. A two-alternative forced-choice procedure was used in conjunction with one of three adaptive psychophysical techniques. These techniques comprised two staircase techniques targeting either 70.7% or 79.4% correct detection (Staircase-71 and Staircase-79), and parameter estimation by sequential testing (PEST), targeting 80% correct detection. Listeners were provided with a rationale for maintaining concentration at weak signal levels. Similar threshold values were obtained from the Staircase-79 and PEST groups in equal numbers of trials. The degree of oscillation in the level of the signal around the value finally chosen as the threshold was comparable for both staircase techniques. Subsequent fixed-level testing did not provide a true indication of the subjects' capabilities. The amount by which percent correct in fixed-level testing differed from expectations based on adaptive testing varied among the techniques. Additional thresholds were obtained in a second session from 30 of the original subjects. Thresholds with both staircase techniques improved by about 1 dB on retest, while thresholds with PEST were constant across sessions. The variability of the data compared well with that from studies involving experienced listeners.

Evoked Potentials

A computer-automated, threshold-tracking procedure was developed for measuring evoked response thresholds on-line. The procedure utilizes an objective detection criterion (correlation) to evaluate whether a response is absent or present. Sound intensity is adjusted according to a modified PEST (parameter estimation by sequential testing) procedure in order to estimate threshold. The computer-automated procedures were used to obtain objective estimates of evoked response thresholds in normal and hearing-impaired chinchillas. The evoked response waveforms stored in the computer were also used to obtain visual judgments of threshold. The objective thresholds determined by the computer were virtually identical to the visual detection thresholds. Thus, the computer-automated procedure provides a reliable, objective, and efficient method of estimating evoked response thresholds.

Within‐subject repeatability of high‐frequency thresholds

Reliability of high‐frequency thresholds was studied in normal‐hearing listeners ranging in age from 20–48 years. Using the PEST (Parameter Estimation by Sequential Testing) procedure, coupler‐referenced thresholds were measured in ten subjects at frequencies 1, 2, 4, 8, 10, 12, and 14 kHz using circumaural earphones. Ten replicates of the measurements were made in all subjects over several weeks. The results indicate a general tendency for standard deviations of thresholds to increase as a function of stimulus frequency in all subjects. Nevertheless, the standard deviations did not exceed 4.0 dB in any of the subjects, suggesting high‐frequency thresholds can be reliably obtained within subject, at least up to 14 kHz. Our findings are in disagreement with earlier results [E. A. G. Shaw, J. Acoust. Soc. Am. 46, 1502–1514 (1969)] which indicated that threshold variability is disproportionately high above 12 kHz in comparison with that below 12 kHz. [Work supported by the EENT Foundation of Louisiana.]

Microcomputer-based estimation of psychophysical thresholds: The Best PEST

A new, maximally efficient technique for measuring psychophysical thresholds (Pentland, 1980) has been implemented on the microcomputer. This PEST (parameter estimation by sequential testing) technique is the most efficient sequential parameter estimation technique possible, given that the form of the psychometric function is known. The technique is similar to but faster and more accurate than other staircase procedures and may be applied whenever staircase techniques are applicable. The “Best PEST” is easily implemented on the micro-computer; a BASIC program for the Apple II which does so is presented. The Best PEST is compared with other staircase procedures, including one recently implemented on a micro-computer (Corwin, Kintz, & Beaty, 1979).