Dissociation Between Speech and Emotion Effects in Short-Term Memory: A Data Reanalysis.

Performance in visual serial recall tasks is often impaired by irrelevant auditory distracters. The duplex-mechanism account of auditory distraction states that if the distracters provide order cues, these interfere with the processing of the order cues in the serial recall task (interference by process). In contrast, the unitary account states that distracters capture only attention on a general level (attentional distraction) without interfering specifically withorder processing. Marsh et al. (2018, Journal of Experimental Psychology-Learning Memory and Cognition, 44, 882-897) reported finding a dissociation between the effects of serial recall tasks and those of a missing-item task on the disruptive effects of speech and of emotional words, as predicted by the duplex-mechanism account. Critically, the reported analyses did not test specifically for the claimed dissociation. Therefore, I reanalyzed the Marsh et al. data and conducted the appropriate analyses. I also tested the dissociation more directly and added a Bayesian hypothesis test to measure the strength of the evidence for a dissociation. Results provided strong evidence for a dissociation (i.e., crossover interaction) between effects of speech and of emotion. Because the duplex-mechanism account predicts this dissociation between speech effects (interference by process) and emotion effects (attentionaldiversion) whereas the unitary account does not, Marsh et al.’s data support the duplex-mechanism account. However, to show that this dissociation is robust, researchers are advised to replicate this dissociation in an adversarial registered report.

Dissociation Between Speech and Emotion Effects in Short-Term Memory: A Data Reanalysis

Performance in visual serial recall tasks is often impaired by irrelevant auditory distracters. The duplex-mechanism account of auditory distraction states that if the distracters provide order cues, these interfere with the processing of the order cues in the serial recall task (interference by process). In contrast, the unitary account states that distracters capture only attention on a general level (attentional distraction) without interfering specifically with order processing. Marsh et al. (2018, Journal of Experimental Psychology-Learning Memory and Cognition, 44, 882-897) reported finding a dissociation between the effects of serial recall tasks and those of a missing-item task on the disruptive effects of speech and of emotional words, as predicted by the duplex-mechanism account. Critically, the reported analyses did not test specifically for the claimed dissociation. Therefore, I reanalyzed the Marsh et al. data and conducted the appropriate analyses. I also tested the dissociation more directly and added a Bayesian hypothesis test to measure the strength of the evidence for a dissociation. Results provided strong evidence for a dissociation (i.e., crossover interaction) between effects of speech and of emotion. Because the duplex-mechanism account predicts this dissociation between speech effects (interference by process) and emotion effects (attentional diversion) whereas the unitary account does not, Marsh et al.’s data support the duplex-mechanism account. However, to show that this dissociation is robust, researchers are advised to replicate this dissociation in an adversarial registered report.

A Bayesian Perspective on Severity: Risky Predictions and Specific Hypotheses

A tradition that goes back to Karl R. Popper assesses the value of a statistical test primarily by its severity: was it a honest and stringent attempt to prove the theory wrong? For "error statisticians" such as Deborah Mayo (1996, 2018), and frequentists more generally, severity is a key virtue in hypothesis tests. Conversely, failure to incorporate severity into statistical inference, as it allegedly happens in Bayesian inference, counts as a major methodological shortcoming. Our paper pursues a double goal: First, we argue that the error-statistical explication of severity has substantive drawbacks (i.e., neglect of research context; lack of connection to specificity of predictions; problematic similarity of degrees of severity to one-sided p-values). Second, we argue that severity matters for Bayesian inference via the value of specific, risky predictions: severity boosts the expected evidential value of a Bayesian hypothesis test. We illustrate severity-based reasoning in Bayesian statistics by means of a practical example and discuss its advantages and potential drawbacks.

How to use and report Bayesian hypothesis tests

This article provides guidance on interpreting and reporting Bayesian hypothesis tests, in order to aid their understanding. To use and report a Bayesian hypothesis test, predicted effect sizes must be specified. The paper will provide guidance in specifying effect sizes of interest (which also will be of relevance to those using frequentist statistics). First, if a minimally interesting effect size can be specified, a null interval is defined as the effects smaller in magnitude than the minimally interesting effect. Then the proportion of the posterior distribution that falls in the null interval indicates the plausibility of the null interval hypothesis. Second, if a rough scale of effect can be determined, a Bayes factor can indicate evidence for a model representing that scale of effect versus a model of H0. Both methods allow data to count against a theory that predicts a difference. By contrast, non-significance does not count against such a theory. Various examples are provided including the suitability of Bayesian analyses for demonstrating the absence of conscious perception under putative subliminal conditions, and its presence in supraliminal conditions.

Conserved novel ORFs in the mitochondrial genome of the ctenophore Beroe forskalii

To date, five ctenophore species’ mitochondrial genomes have been sequenced, and each contains open reading frames (ORFs) that if translated have no identifiable orthologs. ORFs with no identifiable orthologs are called unidentified reading frames (URFs). If truly protein-coding, ctenophore mitochondrial URFs represent a little understood path in early-diverging metazoan mitochondrial evolution and metabolism. We sequenced and annotated the mitochondrial genomes of three individuals of the beroid ctenophore Beroe forskalii and found that in addition to sharing the same canonical mitochondrial genes as other ctenophores, the B. forskalii mitochondrial genome contains two URFs. These URFs are conserved among the three individuals but not found in other sequenced species. We developed computational tools called pauvre and cuttlery to determine the likelihood that URFs are protein coding. There is evidence that the two URFs are under negative selection, and a novel Bayesian hypothesis test of trinucleotide frequency shows that the URFs are more similar to known coding genes than noncoding intergenic sequence. Protein structure and function prediction of all ctenophore URFs suggests that they all code for transmembrane transport proteins. These findings, along with the presence of URFs in other sequenced ctenophore mitochondrial genomes, suggest that ctenophores may have uncharacterized transmembrane proteins present in their mitochondria.

The conditionality principle in high-dimensional regression

Summary Consider a high-dimensional linear regression problem, where the number of covariates is larger than the number of observations and the interest is in estimating the conditional variance of the response variable given the covariates. A conditional and an unconditional framework are considered, where conditioning is with respect to the covariates, which are ancillary to the parameter of interest. In recent papers, a consistent estimator was developed in the unconditional framework when the marginal distribution of the covariates is normal with known mean and variance. In the present work, a certain Bayesian hypothesis test is formulated under the conditional framework, and it is shown that the Bayes risk is a constant. This implies that no consistent estimator exists in the conditional framework. However, when the marginal distribution of the covariates is normal, the conditional error of the above consistent estimator converges to zero, with probability converging to one. It follows that even in the conditional setting, information about the marginal distribution of an ancillary statistic may have a significant impact on statistical inference. The practical implication in the context of high-dimensional regression models is that additional observations where only the covariates are given are potentially very useful and should not be ignored. This finding is most relevant to semi-supervised learning problems where covariate information is easy to obtain.

Bayesian reanalysis of null results reported in Medicine: Strong yet variable evidence for the absence of treatment effects

Efficient medical progress requires that we know when a treatment effect is absent. We considered all 207 Original Articles published in the 2015 volume of the New England Journal of Medicine and found that 45 (21.7%) reported a null result for at least one of the primary outcome measures. Unfortunately, standard statistical analyses are unable to quantify the degree to which these null results actually support the null hypothesis. Such quantification is possible, however, by conducting a Bayesian hypothesis test. Here we reanalyzed a subset of 43 null results from 36 articles using a default Bayesian test for contingency tables. This Bayesian reanalysis revealed that, on average, the reported null results provided strong evidence for the absence of an effect. However, the degree of this evidence is variable and cannot be reliably predicted from the p-value. For null results, sample size is a better (albeit imperfect) predictor for the strength of evidence in favor of the null hypothesis. Together, our findings suggest that (a) the reported null results generally correspond to strong evidence in favor of the null hypothesis; (b) a Bayesian hypothesis test can provide additional information to assist the interpretation of null results.

A phase II trial of gemcitabine (G), cisplatin (C), and nab-paclitaxel (N) in advanced biliary tract cancers (aBTCs): Updated survival analysis.

350 Background: BTCs are often diagnosed at an advanced stage and have a poor prognosis. The standard therapy for aBTCs is the combination of GC. However, the median overall survival (mOS) is dismal at 11.7 months (mos) with a median progression free survival (mPFS) of 8 mos. Methods: A single arm, phase II study was conducted at MD Anderson and Mayo Clinic Arizona. Patients (pts) with aBTC were treated at initial dose level of G/C/N (in mg/m2) at 1000/25/125 (n = 30) which was reduced to lower doses due to grade 3/4 hematological (heme) toxicity (tox) - G/C/N: 800/25/100 (n = 30). Cycles were q21 days with restaging q3 cycles until progression. PFS was the primary endpoint (endpt). Using a Bayesian hypothesis test-based design, we assumed mPFS of 8 mos under the null hypothesis (H0), 10 mos under the alternative (H1). Secondary endpts included mOS, RECIST v1.1 response rate (RR), safety and CA19-9 response. Results: 60 pts were enrolled with 51 being response-evaluable having received more than 1 cycle of therapy (age: median 60 yrs [range 31-77], ECOG PS 0/1 (22/38), M/F (33/27), intrahepatic cholangiocarcinoma/extrahepatic/gallbladder (38/9/13). Median follow-up was 14 mos and median number of treatment (trmt) cycles = 5.24. Pts at initial dose level had significant grade 3/4 heme tox: neutropenia, febrile neutropenia, anemia, and thrombocytopenia leading to trmt discontinuation in 6/30 pts. After dose reduction to G/C/N (in mg/m2) at 800/25/100, trmt was better tolerated with only 3 pts experiencing grade 4 heme tox. Non-heme tox were grade 3 in 19 pts: nausea/vomiting, diarrhea, thromboembolic event/CVA, hypokalemia, constipation, cystitis, LFT elevations. The mPFS = 11.4 mos (95% CI: 6.1, 16.1) and mOS = 19.2 (95%CI: 13.6, NA), 1-year survival rate 67.6%. 51 pts evaluable for response: disease control rate (PR+CR+SD)-84.3% and RR-39%. 12 unresectable cases were operated post trmt with 1 pathologic CR. Conclusions: The combination of GCN was well tolerated at adjusted doses and demonstrates encouraging efficacy having met its mPFS endpt and an impressive mOS higher than historical control. These results merit evaluating GC +/-N in a randomized controlled study. Clinical trial information: NCT02392637.

A phase II trial of gemcitabine (G), cisplatin (C), and nab-paclitaxel (N) in advanced biliary tract cancers (aBTCs).

4018 Background: BTCs are often diagnosed at an advanced stage and have a poor prognosis. The standard therapy for aBTCs is the combination of GC. However, the median overall survival (mOS) is dismal at 11.7 months (mos) with a median progression free survival (mPFS) of 8 mos. Methods: A single arm, phase II study was conducted at MD Anderson and Mayo Clinic Arizona. Patients (pts) with aBTC were treated at initial dose level of G/C/N (in mg/m2) at 1000/25/125 (n = 27) which was reduced to lower doses due to grade 3/4 hematological (heme) toxicity (tox) - G/C/N: 800/25/100 (n = 33). Cycles were q21 days with restaging q3 cycles until progression. PFS was the primary endpoint (endpt). Using a Bayesian hypothesis test-based design, we assumed mPFS of 8 mos under the null hypothesis (H0), 10 mos under the alternative (H1). Secondary endpts included mOS, RECIST v1.1 response rate (RR), safety and CA19-9 response. Results: 60 pts were enrolled with data on 51 available as of the time of this abstract (age: median 60 yrs [range 31-77], ECOG PS 0/1 (17/34), M/F (30/21), intrahepatic cholangiocarcinoma/extrahepatic/gallbladder (32/8/11). Median follow-up was 11.5 mos and median number of treatment (trmt) cycles = 5. Pts at initial dose level had significant grade 3/4 heme tox: neutropenia, febrile neutropenia, anemia, and thrombocytopenia leading to trmt discontinuation in 6/27 pts. After dose reduction to G/C/N (in mg/m2) at 800/25/100, trmt was better tolerated with only 4 pts experiencing grade 4 heme tox. Non-heme tox were grade 3 in 10 pts: nausea/vomiting, diarrhea, thromboembolic event/CVA, hypokalemia, constipation, cystitis, LFT elevations. In the initial 51 pts, mPFS = 11.4 mos (95% CI: 6.1, not reached) and mOS not reached (estimated > 20 mos, 1-year survival rate 66.7%; 95%CI: 65.9-92.2%). 34 pts evaluable for response: disease control rate (PR+CR+SD)-82.3% and RR-32.3%. 3 unresectable cases were operated post trmt with 1 pathologic CR. Conclusions: The combination of GCN was well tolerated at adjusted doses and demonstrates encouraging preliminary efficacy having met its mPFS endpt and a 1-year survival rate higher than historical control. These results merit evaluating GC +/-N in a randomized controlled study. Clinical trial information: NCT02392637.

Marginal likelihood estimation of negative binomial parameters with applications to RNA-seq data

SUMMARY RNA-Seq data characteristically exhibits large variances, which need to be appropriately accounted for in any proposed model. We first explore the effects of this variability on the maximum likelihood estimator (MLE) of the dispersion parameter of the negative binomial distribution, and propose instead to use an estimator obtained via maximization of the marginal likelihood in a conjugate Bayesian framework. We show, via simulation studies, that the marginal MLE can better control this variation and produce a more stable and reliable estimator. We then formulate a conjugate Bayesian hierarchical model, and use this new estimator to propose a Bayesian hypothesis test to detect differentially expressed genes in RNA-Seq data. We use numerical studies to show that our much simpler approach is competitive with other negative binomial based procedures, and we use a real data set to illustrate the implementation and flexibility of the procedure.