Increase In Prediction Error Research Articles

The sound of speed: How grunting affects opponents' anticipation in tennis.

Grunting in tennis is a widespread phenomenon and whether it influences opponents’ predictions of ball trajectory—and if so, why—is subject of ongoing debate. Two alternative hypotheses have been proposed to explain why grunting may impede opponents’ predictions, referred to as the distraction account (i.e., grunts capture attentional resources necessary for anticipation) and the multisensory integration account (i.e., auditory information from the grunt systematically influences ball trajectory prediction typically assumed to rely on visual information). To put these competing hypotheses to test, in the current study we presented tennis players with a series of temporally occluded video clips of tennis rallies featuring experimentally amplified, attenuated, or muted grunting sounds. Participants were asked to predict the ball landing position. Results indicated that higher grunt intensities yielded judgments of longer ball trajectories whereas radial prediction errors were not affected. These results are clearly at odds with the distraction account of grunting, predicting increased prediction errors after higher intensity grunts. In contrast, our findings provide strong support for the multisensory integration account by demonstrating that grunt intensity systematically influences judgments of ball trajectory.

MEBoost: Variable selection in the presence of measurement error.

We present a novel method for variable selection in regression models when covariates are measured with error. The iterative algorithm we propose, Measurement Error Boosting (MEBoost), follows a path defined by estimating equations that correct for covariate measurement error. We illustrate the use of MEBoost in practice by analyzing data from the Box Lunch Study, a clinical trial in nutrition where several variables are based on self-report and, hence, measured with error, where we are interested in performing model selection from a large data set to select variables that are related to the number of times a subject binge ate in the last 28 days. Furthermore, we evaluated our method and compared its performance to the recently proposed Convex Conditioned Lasso and to the "naive" Lasso, which does not correct for measurement error through a simulation study. Increasing the degree of measurement error increased prediction error and decreased the probability of accurate covariate selection, but this loss of accuracy occurred to a lesser degree when using MEBoost. Through simulations, we also make a case for the consistency of the model selected.

A neurocomputational investigation of reinforcement-based decision making as a candidate latent vulnerability mechanism in maltreated children.

Alterations in reinforcement-based decision making may be associated with increased psychiatric vulnerability in children who have experienced maltreatment. A probabilistic passive avoidance task and a model-based functional magnetic resonance imaging analytic approach were implemented to assess the neurocomputational components underlying decision making: (a) reinforcement expectancies (the representation of the outcomes associated with a stimulus) and (b) prediction error signaling (the ability to detect the differences between expected and actual outcomes). There were three main findings. First, the maltreated group (n = 18; mean age = 13), relative to nonmaltreated peers (n = 19; mean age = 13), showed decreased activity during expected value processing in a widespread network commonly associated with reinforcement expectancies representation, including the striatum (especially the caudate), the orbitofrontal cortex, and medial temporal structures including the hippocampus and insula. Second, consistent with previously reported hyperresponsiveness to negative cues in the context of childhood abuse, the maltreated group showed increased prediction error signaling in the middle cingulate gyrus, somatosensory cortex, superior temporal gyrus, and thalamus. Third, the maltreated group showed increased activity in frontodorsal regions and in the putamen during expected value representation. These findings suggest that early adverse environments disrupt the development of decision-making processes, which in turn may compromise psychosocial functioning in ways that increase latent vulnerability to psychiatric disorder.

Interrelation of attention and prediction in visual processing: Effects of task-relevance and stimulus probability

The potentially interactive influence of attention and prediction was investigated by measuring event-related potentials (ERPs) in a spatial cueing task with attention (task-relevant) and prediction (probabilistic) cues. We identified distinct processing stages of this interactive influence. Firstly, in line with the attentional gain hypothesis, a larger amplitude response of the contralateral N1, and Nd1 for attended gratings was observed. Secondly, conforming to the attenuation-by-prediction hypothesis, a smaller negativity in the time window directly following the peak of the N1 component for predicted compared to unpredicted gratings was observed. In line with the hypothesis that attention and prediction interface, unpredicted/unattended stimuli elicited a larger negativity at central-parietal sites, presumably reflecting an increased prediction error signal. Thirdly, larger P3 responses to unpredicted stimuli pointed to the updating of an internal model. Attention and prediction can be considered as differentiated mechanisms that may interact at different processing stages to optimise perception.

Compound Stimulus Presentation Does Not Deepen Extinction in Human Causal Learning.

Models of associative learning have proposed that cue-outcome learning critically depends on the degree of prediction error encountered during training. Two experiments examined the role of error-driven extinction learning in a human causal learning task. Target cues underwent extinction in the presence of additional cues, which differed in the degree to which they predicted the outcome, thereby manipulating outcome expectancy and, in the absence of any change in reinforcement, prediction error. These prediction error manipulations have each been shown to modulate extinction learning in aversive conditioning studies. While both manipulations resulted in increased prediction error during training, neither enhanced extinction in the present human learning task (one manipulation resulted in less extinction at test). The results are discussed with reference to the types of associations that are regulated by prediction error, the types of error terms involved in their regulation, and how these interact with parameters involved in training.

Methane emissions from lactating and non-lactating dairy cows and growing cattle fed fresh pasture

Currently, a fixed methane (CH4) emission factor is used for calculating total CH4 emissions from cattle in the national greenhouse gas inventory of New Zealand, independent of diet composition, cattle class (beef, dairy) or physiological state (growing, lactating, non-lactating). The objectives of this study were to determine CH4 emissions from lactating and non-lactating dairy cows (118 dairy cows; 81 lactating and 37 non-lactating, over 10 periods) and growing dairy heifers (12 measured twice) fed 100% fresh pasture forage in respiration chambers, which in combination with the published data of beef cattle (36 measured twice) fed fresh pasture were used to determine the relationship between CH4 emissions and dry matter intake (DMI), feed quality, cattle class (dairy vs beef) and physiological state (lactating, non-lactating and growing). Before regression analysis the dominant variables (DMI, CH4) needed to be transformed using natural logarithms (Ln) to make the variation in CH4 emissions more homogeneous across the range of data (i.e. stabilise the variance). Over all periods, average DMI ranged from 3.1 to 13.9 kg/day, average CH4 production from 64 to 325 g/day and average CH4 yield from 21.4 to 26.5 g/kg DMI. The DMI alone explained 90.8% of the variation in CH4 production (LnCH4 (g/day) = 3.250 + 0.9487 × LnDMI). Regression was improved to a minor extent (<3%, with associated increased prediction error) by including physiological status, cattle class or dietary composition in the model, in addition to LnDMI, on LnCH4 production. In conclusion, DMI alone was the strongest predictor for CH4 emissions from cattle fed fresh pasture with minor but irrelevant improvements in the prediction when considering pasture quality, cattle class or physiological status.

An Evaluation of Spearman's Hypothesis by Manipulating g Saturation

Spearman's Hypothesis holds that the magnitude of mean White–Black differences on cognitive tests covaries with the extent to which a test is saturated with g. This paper evaluates Spearman's Hypothesis by manipulating the g saturation of cognitive composites. Using a sample of 16,384 people from the General Aptitude Test Battery database, we show that one can decrease mean racial differences in a g test by altering the g saturation of the measure. Consistent with Spearman's Hypothesis, the g saturation of a test is positively and strongly related to the magnitude of White–Black mean racial differences in test scores. We demonstrate that the reduction in mean racial differences accomplished by reducing the g saturation in a measure is obtained at the cost of lower validity and increased prediction errors. We recommend that g tests varying in mean racial differences be examined to determine if the Spearman's Hypothesis is a viable explanation for the results.

The relationship between water and air temperature in chironomid-based paleoclimate reconstructions: Records from boreal and subarctic Finland

Fossil remains of larval chironomids are one of the most useful paleolimnological proxies to reconstruct Holocene paleoclimate. We investigate the within-lake distribution of chironomids in relation to water temperature and test the usability of intralake calibration in chironomid-based temperature reconstructions. We compare our findings against the traditional multilake calibration technique in Holocene sediment cores from Finland. The results show that intralake water temperature optima of common taxa correspond closely with mean July air temperature optima in a regional multilake dataset, with exception of a few vegetation-associated taxa. A calibration model based on intralake water temperatures shows performance statistics that correspond to the general performance statistics level of multilake calibration models. However, the intralake model has somewhat increased prediction error. Despite the fact that the intralake model is sensitive to poor modern analogues, it shows correlation with the results using the multilake calibration approach in the Holocene reconstructions. Our results confirm that the within-lake distribution of chironomids along a water depth gradient is closely linked with water temperature gradient. This provides evidence for the close relationship between chironomids and temperature in general and suggests that, in addition to the regional scale, water temperature is a key variable in determining chironomid distribution at the site-specific scale. However, long-term changes in intralake gradients in other depth-related factors, such as hypolimnetic oxygen conditions, extent of photic zone and habitat distribution, may hamper the quantitative accuracy of chironomid-based temperature reconstructions, though the results also suggest that these changes are likely to have little general impact on the trends of chironomid-based air temperature reconstructions.

The right hemisphere supports but does not replace left hemisphere auditory function in patients with persisting aphasia

In this study, we used magnetoencephalography and a mismatch paradigm to investigate speech processing in stroke patients with auditory comprehension deficits and age-matched control subjects. We probed connectivity within and between the two temporal lobes in response to phonemic (different word) and acoustic (same word) oddballs using dynamic causal modelling. We found stronger modulation of self-connections as a function of phonemic differences for control subjects versus aphasics in left primary auditory cortex and bilateral superior temporal gyrus. The patients showed stronger modulation of connections from right primary auditory cortex to right superior temporal gyrus (feed-forward) and from left primary auditory cortex to right primary auditory cortex (interhemispheric). This differential connectivity can be explained on the basis of a predictive coding theory which suggests increased prediction error and decreased sensitivity to phonemic boundaries in the aphasics' speech network in both hemispheres. Within the aphasics, we also found behavioural correlates with connection strengths: a negative correlation between phonemic perception and an inter-hemispheric connection (left superior temporal gyrus to right superior temporal gyrus), and positive correlation between semantic performance and a feedback connection (right superior temporal gyrus to right primary auditory cortex). Our results suggest that aphasics with impaired speech comprehension have less veridical speech representations in both temporal lobes, and rely more on the right hemisphere auditory regions, particularly right superior temporal gyrus, for processing speech. Despite this presumed compensatory shift in network connectivity, the patients remain significantly impaired.

The thing that should not be: predictive coding and the uncanny valley in perceiving human and humanoid robot actions

Using functional magnetic resonance imaging (fMRI) repetition suppression, we explored the selectivity of the human action perception system (APS), which consists of temporal, parietal and frontal areas, for the appearance and/or motion of the perceived agent. Participants watched body movements of a human (biological appearance and movement), a robot (mechanical appearance and movement) or an android (biological appearance, mechanical movement). With the exception of extrastriate body area, which showed more suppression for human like appearance, the APS was not selective for appearance or motion per se. Instead, distinctive responses were found to the mismatch between appearance and motion: whereas suppression effects for the human and robot were similar to each other, they were stronger for the android, notably in bilateral anterior intraparietal sulcus, a key node in the APS. These results could reflect increased prediction error as the brain negotiates an agent that appears human, but does not move biologically, and help explain the ‘uncanny valley’ phenomenon.

Determination of Basic Density and Moisture Content of Merchantable Loblolly Pine Logs by near Infrared Spectroscopy

The use of near infrared (NIR) diffuse reflectance spectroscopy for the estimation of basic density and moisture content of loblolly pine ( Pinus taeda L.) logs was explored. A total of 219 wood disks representing the large and small ends of 146 merchantable logs 4.87 m in length, ranging in age from 13 years to 19 years, were analysed. NIR spectra were obtained from the transverse surface of the disks when the wood was green using a Foss XDS NIR System coupled with a Smart Probe Analyser. Each disk was divided into four quadrants and two spectra were collected per quadrant (one representing juvenile wood and one mature wood). NIR calibration models were fitted using pre-processed spectra with multiplicative scatter correction and the dimensionality of the models was kept constant for comparison purposes. Successful models based on an average of all spectra collected from both ends of the logs were obtained for basic density ( R2c = 0.84) and moisture content ( R2c = 0.85) using nine factors for both properties. The predictive ability of the calibrations was acceptable, with mean square errors of 22.8 kg m−3 for basic density and 2.1% for moisture content. It was found that the number of spectra per disk could be reduced to four readings but further reductions increased prediction errors considerably. Calibration models based only on spectra collected from one of the log ends were also examined. Using large-end diameter spectra only, the fit statistics for basic density models fell to R2 c=0.76 for (eight spectra) and R2c=0.71 (four spectra). Similarly, the fit statistics for moisture content calibrations dropped to R2c=0.76 for (eight spectra) and R2c=0.72 (four spectra). Calibrations based only on small-end diameter spectra were considerably weaker than those based on large-end diameter spectra.

Psychophysiological characteristics of autism in children

The potentially interactive influence of attention and prediction was investigated by measuring event-related potentials (ERPs) in a spatial cueing task with attention (task-relevant) and prediction (probabilistic) cues. We identified distinct processing stages of this interactive influence. Firstly, in line with the attentional gain hypothesis, a larger amplitude response of the contralateral N1, and Nd1 for attended gratings was observed. Secondly, conforming to the attenuation-by-prediction hypothesis, a smaller negativity in the time window directly following the peak of the N1 component for predicted compared to unpredicted gratings was observed. In line with the hypothesis that attention and prediction interface, unpredicted/unattended stimuli elicited a larger negativity at central-parietal sites, presumably reflecting an increased prediction error signal. Thirdly, larger P3 responses to unpredicted stimuli pointed to the updating of an internal model. Attention and prediction can be considered as differentiated mechanisms that may interact at different processing stages to optimise perception.

The role of the dominant descriptor in targeted quantitative structure property relationships

The targeted quantitative structure–property relationship (TQSPR) method of Brauner et al. [2006. A structurally “targeted” QSPR method for property prediction. Industrial & Engineering Chemistry Research 45, 8430–8437] is analyzed in this study with examples involving prediction of critical temperatures of hydrocarbons. Targeted are 108 hydrocarbons, for which experimental critical temperature values can be obtained from published sources. Each hydrocarbon structure is represented by 1630 calculated values of molecular descriptors. The availability of a wide variety of molecular descriptors and small training sets comprising compounds structurally similar to the target enables the detection of one, dominant descriptor, which is well correlated with a particular property and is sufficient to serve as a single predictor for this property. The analysis of the relationship between the dominant descriptor and property values for the training set and the target compound enables better understanding of the main causes for increased prediction errors. Based on this analysis, new quantitative indicators for assessing the reliability of the predicted property value are proposed, which may reduce considerably the uncertainty associated with property prediction.

Enhancing calibration models for non-invasive near-infrared spectroscopical blood glucose determination

Partial least-squares regression (PLS) and radial basis function (RBF) networks are used to compute calibration models for non-invasive blood glucose determination by NIR diffuse reflectance spectroscopy. A model computation shows that even extremely small deviations of the spectra induce increased prediction errors. Since the spectral contribution of blood glucose is much smaller than deviations resulting from the non-invasive measuring process a method based on Pearson’s correlation coefficient can be used for evaluating the quality of the recorded spectra during the prediction step. Another method is based on the leverage values from the hat matrix of the RBF network. Both methods lead to a significant decrease in prediction error.

DC matching for motion estimation

A new block matching algorithm is presented which saves computation time by using a simple and effective DC matching technique. Preliminary results indicate that this algorithm reduces the computation complexity by a factor of 6.5 at the expense of 0 to 0.9% increased prediction error as compared to the conventional full-search method.

Separating the ground and airborne laser sampling phases to estimate tropical forest basal area, volume, and biomass

Airborne laser profiling data were used to estimate the basal area, volume, and biomass of primary tropical forests. A procedure was developed and tested to divorce the laser and ground data collection efforts using three distinct data sets acquired in and over the tropical forests of Costa Rica. Fixed-area ground plot data were used to simulate the height characteristics of the tropical forest canopy and to simulate laser measurements of that canopy. On two of the three study sites, the airborne laser estimates of basal area, volume, and biomass grossly misrepresented ground estimates of same. On the third study site, where the widest ground plots were utilized, airborne and ground estimates agreed within 24%. Basal area, volume, and biomass prediction inaccuracies in the first two study areas were tied directly to disagreements between simulated laser estimates and the corresponding airborne measurements of average canopy height, height variability, and canopy density. A number of sampling issues were investigated; the following results were noted in the analyses of the three study areas. 1) Of the four ground segment lengths considered (25 m, 50 m, 75 m, and 100 m), the 25 m segment length introduced a level of variability which may severely degrade prediction accuracy in these Costa Rican primary tropical forests. This effect was more pronounced as plot width decreased. A minimum segment length was on the order of 50 m. 2) The decision to transform or not to transform the dependent variable (e.g., biomass) was by far the most important factor of those considered in this experiment. The natural log transformation of the dependent variable increased prediction error, and error increased dramatically at the shorter segment lengths. The most accurate models were multiple linear models with forced zero intercept and an untransformed dependent variable. 3) General linear models were developed to predict basal area, volume, and biomass using airborne laser height measurements. Useful laser measurements include average canopy height, all pulses (h̄a), average canopy height, canopy hits (h̄c) and the coefficients of variation of these terms (ca and cc). Coefficients of determination range from 0.4 to 0.6. Based on this research, airborne laser and ground sampling procedures are proposed for use for reconnaissance level surveys of inaccessible forested regions.

An improved deadbeat rectifier regulator using a neural net predictor

This paper proposes a new input current reference prediction scheme for the deadbeat control of a three-phase rectifier used in AC/DC/AC converters. The inherent lag in deadbeat control is compensated by predicting the reference resulting in better performance. The proposed predictor consists of a neural net which is trained on-line and predicts the slow varying and periodic trends of the current reference plus a linear first order predictor which predicts the fast variations of the current reference time signal. A CRITIC decides if the neural net training is sufficient and therefore whether or not to use the prediction in the control loop. The learning rule used allows neural net weights to be trained whenever a parameter change causes an increased prediction error. This predictive-regulator is shown to result in improved performance in steady state, in the presence of input voltage imbalance or load variations.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Prediction Error and Its Estimation for Subset-Selected Models

Strategies are compared for development of a linear regression model and the subsequent assessment of its predictive ability. Simulations were performed as a designed experiment over a range of data structures. Approaches using a forward selection of variables resulted in slightly smaller prediction errors and less biased estimators of predictive accuracy than all possible subsets selection but often did not improve on the full model. Random and balanced data splitting resulted in increased prediction errors and estimators with large mean squared error. To maximize predictive accuracy while retaining a reliable estimate of that accuracy, it is recommended that the entire sample usually he used for model development and assessment.

Prediction Error and Its Estimation for Subset-Selected Models

Strategies are compared for development of a linear regression model and the subsequent assessment of its predictive ability. Simulations were performed as a designed experiment over a range of data structures. Approaches using a forward selection of variables resulted in slightly smaller prediction errors and less biased estimators of predictive accuracy than all possible subsets selection but often did not improve on the full model. Random and balanced data splitting resulted in increased prediction errors and estimators with large mean squared error. To maximize predictive accuracy while retaining a reliable estimate of that accuracy, it is recommended that the entire sample usually he used for model development and assessment.

Contemporary Groups for Genetic Evaluations

Contemporary groups are used to remove biases from genetic evaluations due to differential effects such as management associated with the grouping. Numerous groups, however, can result in small numbers of records per subclass with associated loss of effective number of daughters for sire evaluation and increased prediction error variance. Thus, in practice, mean square error, bias squared plus prediction error variance, may be more meaningful than bias alone or prediction error variance. Considering contemporary groups as fixed removes bias due to association between effects corresponding to contemporary groups and sires. If contemporary groups are considered random, then effective number of daughters is increased at the expense of possible bias. Various compromises may be effective for increasing genetic gain. Arbitrary definition of contemporary groups can include herd-year-season of freshening, lactation number, registered or nonregistered, sampling or postsampling daughters, and special treatments among others. The assumption of homogeneous genetic and residual variances is likely to be incorrect. Alternative methods include simple transformations, a two-step transformation, and multiple trait modeling. Multiple trait analyses may include the assumption of genetic correlations of unity, common genetic and heterogeneous residual variances, and joint estimation of genetic values and variances.