Interpretation Biases in Pain: Validation of Two New Stimulus Sets

Adults with chronic pain interpret ambiguous information in a pain and illness related fashion. However, limitations have been highlighted with traditional experimental paradigms used to measure interpretation biases. Whilst ambiguous scenarios have been developed to measure interpretation biases in adolescents with pain, no scenario sets exist for use with adults. Therefore, the present study: (i) sought to validate a range of ambiguous scenarios suitable for measuring interpretation biases in adults, whilst also allowing for two response formats (forced-choice and free response); and (ii) investigate paradigm efficacy, by assessing the effects of recent pain experiences on task responding. A novel ambiguous scenarios task was administered to adults (N = 241). Participants were presented with 62 ambiguous scenarios comprising 42 that could be interpreted in a pain/pain-illness or non-pain/non-pain illness manner: and 20 control scenarios. Participants generated their own solutions to each scenario (Word Generation Task), then rated how likely they would be to use two researcher-generated solutions to complete each scenario (Likelihood Ratings Task). Participants also rated their subjective experiences of pain in the last 3 months. Tests of reliability, including inter-rater agreement and internal consistency, produced two ambiguous scenario stimulus sets containing 18 and 20 scenarios, respectively. Further analyses revealed adults who reported more recent pain experiences were more likely to endorse the pain/pain-illness solutions in the Likelihood Ratings Task. This study provides two new stimulus sets for use with adults (including control items) in pain research and/or interventions. Results also provide evidence for a negative endorsement bias in adults.

OptAGAN: Entropy-based Finetuning on Text VAE-GAN

Transfer learning through large pre-trained models has changed the landscape of current applications in natural language processing (NLP). Recently Optimus, a variational autoencoder (VAE) which combines two pre-trained models, BERT and GPT-2, has been released, and its combination with generative adversarial networks (GANs) has been shown to produce novel, yet very human-looking text. The Optimus and GANs combination avoids the troublesome application of GANs to the discrete domain of text, and prevents the exposure bias of standard maximum likelihood methods. We combine the training of GANs in the latent space, with the finetuning of the decoder of Optimus for single word generation. This approach lets us model both the high-level features of the sentences, and the low-level word-by-word generation. We finetune using reinforcement learning (RL) by exploiting the structure of GPT-2 and by adding entropy-based intrinsically motivated rewards to balance between quality and diversity. We benchmark the results of the VAE-GAN model, and show the improvements brought by our RL finetuning on three widely used datasets for text generation, with results that greatly surpass the current state-of-the-art for the quality of the generated texts.

Sequence-Level Training for Non-Autoregressive Neural Machine Translation

In recent years, Neural Machine Translation (NMT) has achieved notable results in various translation tasks. However, the word-by-word generation manner determined by the autoregressive mechanism leads to high translation latency of the NMT and restricts its low-latency applications. Non-Autoregressive Neural Machine Translation (NAT) removes the autoregressive mechanism and achieves significant decoding speedup through generating target words independently and simultaneously. Nevertheless, NAT still takes the word-level cross-entropy loss as the training objective, which is not optimal because the output of NAT cannot be properly evaluated due to the multimodality problem. In this article, we propose using sequence-level training objectives to train NAT models, which evaluate the NAT outputs as a whole and correlates well with the real translation quality. Firstly, we propose training NAT models to optimize sequence-level evaluation metrics (e.g., BLEU) based on several novel reinforcement algorithms customized for NAT, which outperforms the conventional method by reducing the variance of gradient estimation. Secondly, we introduce a novel training objective for NAT models, which aims to minimize the Bag-of-Ngrams (BoN) difference between the model output and the reference sentence. The BoN training objective is differentiable and can be calculated efficiently without doing any approximations. Finally, we apply a three-stage training strategy to combine these two methods to train the NAT model.We validate our approach on four translation tasks (WMT14 En↔De, WMT16 En↔Ro), which shows that our approach largely outperforms NAT baselines and achieves remarkable performance on all translation tasks. The source code is available at https://github.com/ictnlp/Seq-NAT.

Narrative thinking lingers in spontaneous thought

What we think about at any moment is shaped by what preceded it. Why do some experiences, such as reading an immersive story, feel as if they linger in mind beyond their conclusion? In this study, we hypothesize that the stream of our thinking is especially affected by "deeper" forms of processing, emphasizing the meaning and implications of a stimulus rather than its immediate physical properties or low-level semantics (e.g., reading a story vs. reading disconnected words). To test this idea, we presented participants with short stories that preserved different levels of coherence (word-level, sentence-level, or intact narrative), and we measured participants’ self-reports of lingering and spontaneous word generation. Participants reported that stories lingered in their minds after reading, but this effect was greatly reduced when the same words were read with sentence or word-order randomly shuffled. Furthermore, the words that participants spontaneously generated after reading shared semantic meaning with the story’s central themes, particularly when the story was coherent (i.e., intact). Crucially, regardless of the objective coherence of what each participant read, lingering was strongest amongst participants who reported being ‘transported’ into the world of the story while reading. We further generalized this result to a non-narrative stimulus, finding that participants reported lingering after reading a list of words, especially when they had sought an underlying narrative or theme across words. We conclude that recent experiences are most likely to exert a lasting mental context when we seek to extract and represent their deep situation-level meaning.

Vocal features obtained through automated methods in verbal fluency tasks can aid the identification of mixed episodes in bipolar disorder

There is a lack of consensus on the diagnostic thresholds that could improve the detection accuracy of bipolar mixed episodes in clinical settings. Some studies have shown that voice features could be reliable biomarkers of manic and depressive episodes compared to euthymic states, but none thus far have investigated whether they could aid the distinction between mixed and non-mixed acute bipolar episodes. Here we investigated whether vocal features acquired via verbal fluency tasks could accurately classify mixed states in bipolar disorder using machine learning methods. Fifty-six patients with bipolar disorder were recruited during an acute episode (19 hypomanic, 8 mixed hypomanic, 17 with mixed depression, 12 with depression). Nine different trials belonging to four conditions of verbal fluency tasks—letter, semantic, free word generation, and associational fluency—were administered. Spectral and prosodic features in three conditions were selected for the classification algorithm. Using the leave-one-subject-out (LOSO) strategy to train the classifier, we calculated the accuracy rate, the F1 score, and the Matthews correlation coefficient (MCC). For depression versus mixed depression, the accuracy and F1 scores were high, i.e., respectively 0.83 and 0.86, and the MCC was of 0.64. For hypomania versus mixed hypomania, accuracy and F1 scores were also high, i.e., 0.86 and 0.75, respectively, and the MCC was of 0.57. Given the high rates of correctly classified subjects, vocal features quickly acquired via verbal fluency tasks seem to be reliable biomarkers that could be easily implemented in clinical settings to improve diagnostic accuracy.

Cerebral lateralisation of first and second languages in bilinguals assessed using functional transcranial Doppler ultrasound

Background: Lateralised language processing is a well-established finding in monolinguals. In bilinguals, studies using fMRI have typically found substantial regional overlap between the two languages, though results may be influenced by factors such as proficiency, age of acquisition and exposure to the second language. Few studies have focused specifically on individual differences in brain lateralisation, and those that have suggested reduced lateralisation may characterise representation of the second language (L2) in some bilingual individuals. Methods: In Study 1, we used functional transcranial Doppler sonography (FTCD) to measure cerebral lateralisation in both languages in high proficiency bilinguals who varied in age of acquisition (AoA) of L2. They had German (N = 14) or French (N = 10) as their first language (L1) and English as their second language. FTCD was used to measure task-dependent blood flow velocity changes in the left and right middle cerebral arteries during phonological word generation cued by single letters. Language history measures and handedness were assessed through self-report. Study 2 followed a similar format with 25 Japanese (L1) /English (L2) bilinguals, with proficiency in their second language ranging from basic to advanced, using phonological and semantic word generation tasks with overt speech production. Results: In Study 1, participants were significantly left lateralised for both L1 and L2, with a high correlation (r = .70) in the size of laterality indices for L1 and L2. In Study 2, again there was good agreement between LIs for the two languages (r = .77 for both word generation tasks). There was no evidence in either study of an effect of age of acquisition, though the sample sizes were too small to detect any but large effects. Conclusion: In proficient bilinguals, there is strong concordance for cerebral lateralisation of first and second language as assessed by a verbal fluency task.

Improved Framework using Rider Optimization Algorithm for Precise Image Caption Generation

“Image Captioning is the process of generating a textual description of an image”. It deploys both computer vision and natural language processing for caption generation. However, the majority of the image captioning systems offer unclear depictions regarding the objects like “man”, “woman”, “group of people”, “building”, etc. Hence, this paper intends to develop an intelligent-based image captioning model. The adopted model comprises of few steps like word generation, sentence formation, and caption generation. Initially, the input image is subjected to the Deep learning classifier called Convolutional Neural Network (CNN). Since the classifier is already trained in the relevant words that are related to all images, it can easily classify the associated words of the given image. Further, a set of sentences is formed with the generated words using Long-Short Term Memory (LSTM) model. The likelihood of the formed sentences is computed using the Maximum Likelihood (ML) function, and the sentences with higher probability are taken, which is further used for generating the visual representation of the scene in terms of image caption. As a major novelty, this paper aims to enhance the performance of CNN by optimally tuning its weight and activation function. This paper introduces a new enhanced optimization algorithm Rider with Randomized Bypass and Over-taker update (RR-BOU) for this optimal selection. In the proposed RR-BOU is the enhanced version of the Rider Optimization Algorithm (ROA). Finally, the performance of the proposed captioning model is compared over other conventional models with respect to statistical analysis.

A magnetoencephalography dataset for motor and cognitive imagery-based brain-computer interface

Recent advancements in magnetoencephalography (MEG)-based brain-computer interfaces (BCIs) have shown great potential. However, the performance of current MEG-BCI systems is still inadequate and one of the main reasons for this is the unavailability of open-source MEG-BCI datasets. MEG systems are expensive and hence MEG datasets are not readily available for researchers to develop effective and efficient BCI-related signal processing algorithms. In this work, we release a 306-channel MEG-BCI data recorded at 1KHz sampling frequency during four mental imagery tasks (i.e. hand imagery, feet imagery, subtraction imagery, and word generation imagery). The dataset contains two sessions of MEG recordings performed on separate days from 17 healthy participants using a typical BCI imagery paradigm. The current dataset will be the only publicly available MEG imagery BCI dataset as per our knowledge. The dataset can be used by the scientific community towards the development of novel pattern recognition machine learning methods to detect brain activities related to motor imagery and cognitive imagery tasks using MEG signals.

Exclusion

The theme of exclusion emerged at the IAMCR conference 2019 in Oregon, where we* hosted several panels that dealt with either the welcomed potential of the Web to help excluded groups to be heard and seen (e.g. #metoo online activism), or with negative exclusion and ostracising effects of media—such as hate speech on social media, strategies of “othering” in newspapers or “symbolic annihilation” (Tuchmann) on screens. In their special issue of Media International Australia on the role of media in social inclusion and exclusion in 2012, the editors Jacqui Ewart and Collette Snowden argued that research on exclusion and media was still a highly relevant topic (62). Eight years later we would like to second this claim: in a society where we still encounter inequality, polarisation, hate, and ostracism, the role of media needs to be considered even more than before. The saturation with media of our lifeworlds necessitates a deeper understanding of the mechanisms of exclusion in and through media. We begin this highly topical matter by looking into the past: Mark Dang-Anh’s feature article addresses exclusion and agency as linguistic practice in World War II. The field post letters of German soldiers were subject to control and censorship and soldiers—as part of National Socialism—thus performed discursive practices of inclusion and exclusion. Field post letters are seen in this sense as a dispositif (Foucault): on the one hand as a “rather rigid structure” and on the other hand as “potentially discourse-excluding social stratification of private communication”. Dang-Anh’s analytical thoughts on exclusion and media within a fascist dictatorship allow for a realignment in approaching mechanisms of exclusion in modern, multivocal societies. His differentiation into infrastructural and interactional processes of inclusion is furthermore guiding in conceiving the following articles on exclusion that play out on a structural level and on the level of communication strategies and practices. Moving from past to present, Sarah Baker and Amanda Rutherford reveal the manyfold mechanisms of stereotyping, and hence exclusion, of authentic representations of lesbians and gays in Hollywood films and TV series. The contribution reminds us of the complex and ambiguous inclusion/exclusion nexus of screen representations where the goodwill to create inclusive stories can result in new forms of exclusion. Drawing on The L Word and its sequel The L Word: Generation Q, the authors illustrate how both series successfully denaturalise the hegemonic straight gaze and how The L Word: Generation Q managed to address exclusion in a more nuanced way than its predecessor. Staying within the field of screen representation, Claudia Wegener, Elizabeth Prommer, and Christine Linke set out to investigate and reveal the exclusion of the diversity and variety of female life on YouTube in Germany. In their empirical study, including more than 2,000 videos from YouTubers, the authors’ findings expose and unmask the platform as highly gendered and disadvantageous to girls and women. Women are significantly underrepresented in the most popular videos, they are less visible as content producers, more limited in their range of topics, and they appear predominantly in relation to topics with a stereotypically female connotation. Having also a focus on influencers but applying a transnational and comparative perspective on influencer markets in Australia, Japan, and Korea, Crystal Abidin, Tommaso Barbetta, and Jin Lee describe the transformation of the role of influencers in the advertising market during the COVID-19 pandemic. Influencers, the authors argue, were “primed to be the dominant and default mode of advertising and communication in the post-COVID-19 era”. Yet, in order to remain successful under the changing conditions, new strategies and new formats were necessary. In their account of these developments the authors critically reflect on those able to keep up and those who are left behind in the changing market dynamics. In his theoretical consideration “A Flattering Robocalypse”, Matthew Horrigan turns towards exclusion by design, and one exclusionary practice embedded in the AI systems of the Web, the CAPTCHA. Horrigan identifies CAPTCHA as a system designed to distinguish humans from non-humans while applying a test-game that humans ultimately cannot win because it puts humans, at long last, at a disadvantage. His reflections provide a conceptual entry into the tension between human agency and machine AI which is at the heart of our capacity to act in a digitalised world. The last two articles both address exclusion mechanisms and strategies of particular groups in their use of social media. Julia Stüwe and Juliane Wegener examine over 7,500 “posts” and more than 4,000 “stories” of 142 German-speaking young cancer bloggers. They explore their social media behaviours and how young cancer bloggers use social media—particularly Instagram—for “information sharing, exchanging ideas, networking, and to address their unmet needs of the real world”. Yet, the bloggers exclude illness-related narratives from their posts, thus creating the paradox of self-chosen exclusion. In a self-reflexive study about the use of social media by academics, Franziska Thiele presents her results based on interviews with 16 communication scholars at different stages in their academic career. ResearchGate, the author argues, mainly targets people working in the scientific field, while excluding everyone else. It thus serves as “a symbol of distinction from other groups”. Yet, while clearly supporting careers in a highly competitive environment, Thiele’s study also reveals a reluctance among her participants to use social media in a work-related context. Note * The editors of this issue are the management team of the IAMCR section Mediated Communication, Public Opinion, and Society. References Ewart, Jacqui, and Collette Snowden. "The Media’s Role in Social Inclusion and Exclusion." Media International Australia 142.1 (2012): 61–63. <https://doi.org/10.1177/1329878X1214200108>. Tuchman, Gaye. "The Symbolic Annihilation of Women by the Mass Media." Culture and Politics. Eds. Lane Crothers et al. New York: Palgrave Macmillan, 2000. 150–174. <https://doi.org/10.1007/978-1-349-62397-6_9>.