Centro Universitário Farias Brito: FB UNI Portal de Periódicos
Not a member yet
39872 research outputs found
Sort by
The perceptual primacy of feeling: Affectless machine vision models explain a majority of variance in human visually-evoked affect
No full textLooking at the world often involves not just seeing things, but feeling things. Modern feedforward machine vision systems that learn to perceive the world in the absence of active physiology, deliberative thought, or any form of feedback that resembles human affective experience offer tools to demystify relationship between seeing and feeling, and to assess how much of visually-evoked affective experiences may be a straightforward function of representation learning over natural image statistics. In this work, we deploy a diverse sample of 180 state-of-the-art deep neural network models trained only on canonical computer vision tasks to predict human ratings of arousal, valence, and beauty for images from multiple categories (objects, faces, scenescapes, art) across two datasets. Importantly, we use the features of these models without additional learning, linearly decoding human affective responses from network activity in much the same way neuroscientists decode information from neural recordings. Aggregate analysis across our survey demonstrates that predictions from purely perceptual models explain a majority of the explainable variance in average ratings of arousal, valence, and beauty alike. Finer-grained analysis within our survey (e.g. comparisons between shallower and deeper layers, or between randomly-initialized, category-supervised, and self-supervised models) point to rich, pre-conceptual abstraction (learned from diversity of visual experience) as a key driver of these predictions. Taken together, these results provide further computational evidence for an information-processing account of visually-evoked affect linked directly to efficient representation learning over natural image statistics, and hint at a computational locus of affective and aesthetic valuation immediately proximate to perception
Realizing a Global Survey of Emigrants through Facebook and Instagram
No full text*** Please note that a revised and final version of this contribution has been published as open access article in Comparative Migration Studies. --- Pötzschke, S., & Weiß, B. (2025). Surveying emigrants worldwide – using Facebook and Instagram to recruit respondents in cross-national (e)migration research. Comparative Migration Studies, 13(56). https://doi.org/10.1186/s40878-025-00464-w ***
Research on international migrants has seen a sharp increase during the last decades, yet sampling them remains a major challenge, especially in a cross-national setting and on a global scale. While various sampling methods are established in the field, most of them cannot easily be implemented globally due to their dependence on specific administrative or infrastructure elements or simply their costs. Since Social Networking Sites (SNS) operate on a global scale, they provide a sampling frame that can be utilized for the targeted recruitment of migrants worldwide. Increasingly used for research purposes and among the largest and most popular SNSs are Facebook and Instagram. In our project GEOOS (German Emigrants Overseas Online Survey), we utilize paid advertisements on these networks to target German emigrants, particularly Germans living outside of Europe. Our research aims to ascertain whether such ads could be used to recruit a nonprobability (migrant) sample on a global scale. More specifically, we are interested in the success of this approach concerning three performance indicators: Cost efficiency, coverage, and sample size. Our advertisement campaign ran for 18 days and resulted in total costs of about 2,223 Euro. This investment led a total of 3,895 individuals to complete the survey; of those, 98 percent belonged to the target population, meaning they were (a) either born in Germany or held German citizenship and (b) did not live in Germany. GEOOS participants lived in a total of 148 countries and territories around the globe. Similar to findings reported in previous studies on this target population, the largest sub-groups resided in predominantly Anglo-phone countries; however, taken together, participants in these countries only constitute 38 percent of our overall sample, with nearly a quarter of GEOOS participants (n = 867) living in Middle and South America, 862 residing in Asian countries, and 476 in Africa. Furthermore, a considerable share of our sample is constituted by individuals who would either not have been included in a sampling frame based on German population registers or who would have been unlikely to be reached through this method due to incomplete or outdated information
The upside: How people make sense of difficulty matters in a crisis
No full textWe tested the prediction that how people respond to all-encompassing life difficulties that may require taking on novel difficult tasks or goals is a function of what they infer about their identities from these experiences of difficulty. We focused on the COVID-19 pandemic and identity-based motivation theory to test our predictions (N=698 U.S. adults, three datasets). People were more likely to see silver linings if they endorsed difficulty-as-importance (experienced difficulties with a task/goal as implying its importance) and difficulty-as-improvement (experienced life difficulties as possibly making them better people). Our structural equation models revealed that people who endorsed difficulty-as-importance were more likely to mask, distance, and wash hands in large part because they saw a silver lining for themselves in the pandemic; for difficulty-as-improvement, effects on action were fully mediated by seeing silver linings. Taken together, our results suggest that people apply their difficulty-as-importance and difficulty-as-improvement mindsets to cope with novel life difficulties
A Continuous Measure of Object-Based Attention Sheds New Light on Its Underlying Mechanisms
No full textAttention is largely guided by objects; when attending to a location on an object, observers react faster to targets appearing in other locations within the object than outside the object. While crucial to our understanding of attention, the mechanisms underlying object-based attention remain a topic of debate. Moreover, the behavioral effects of object-based attention are often small and inconsistent, posing challenges for obtaining unequivocal evidence about its mechanisms. To address these challenges, we employ a continuous, response-free measure of object-based attention that utilizes attentional modulation of the pupillary light response (PLR). This method enables continuous tracking of attention across space while avoiding potential inconsistencies introduced by behavioral responses. In each trial, a cue appeared inside or outside a black or white object, followed by a target in the same location as the cue or in a different location. Across four experiments, attentional modulations of the PLR were consistently stronger when attention shifted from a location outside the object to inside it (cue outside, target inside) than in the reverse scenario (cue inside, target outside). This finding reveals a critical difference in the dynamics of attentional shifts involving objects, suggesting that spatial reallocation of attention is more effective when disengagement from an object is not required. These results align with the shifting account of object-based attention, which underscores the higher cost of shifting attention between objects than within an object
Road Not Taken： Why Artificial Intelligence Will Not Re-define Politics
No full textArtificial intelligence is not the only or main force thrown upon politics by current scientific revolution. No man in social sense is not, or more than AI, and rulers in the past has long been mastering the more versatile and volatile Slave Intelligence. As a political community with enough surplus product can easily withstand malgovernance, without necessity or outcry for “Optimal Rule”, artificial intelligence will unlikely outgun or outvote human rulers. Among international competition barbarized by prospect of perpetual world domination through new technology, natural sciences and politics will no longer know ethics. People of nation states, if too little to rule themselves, will ask for a quasi-omnipotent protector, against artificial intelligence and foreign conquerors, at the price of total submission. This protector will be awarded with those very technology for a general artificial intelligence, to become a superhuman utterly integrated with machines, sooner, surer, and quicker than general artificial intelligence, or any other system of production and distribution past or foreign. Yet this will not make it any less human, as such integration is never absent or rare in human history. Through conquest, this superhuman ruler, itself a creation facilitated by international politics, may end international politics altogether, along with nation states. A superhuman ruler born of international politics, technology, and fear, is thus more likely the outcome of current scientific revolution, than scenarios of AI takeover/cataclysm/enthronement. Yet though thinkers of the past has already many foreshadows for such an outcome, actual future is still decided by actions of generations to come
Electoral Competition with Targeted Voting Costs
No full textHow do voting laws impact elections? We highlight how laws targeting a specific group of citizens can have weak effects on turnout and vote shares but substantial effects on policy platforms, thereby influencing substantive representation even if there are no observable effects on participation. To parse these effects, we analyze a model of electoral competition with endogenous turnout and targeted voting costs. Each party anticipates the direct effect of raising one side's voting costs: discouraging targeted citizens from voting. Consequently, both platforms shift towards the untargeted group. These platform adjustments mobilize targeted citizens and demobilize the untargeted, muting the net impact on turnout and vote shares—consistent with empirical evidence for small electoral effects. Policy effects, however, hurt targeted citizens and their aligned party. The targeted group's size amplifies these effects. Our results address party competition, participation, representation, and normative and empirical evaluations of voting laws
A Systematic Evaluation of Wording Effects Modeling Under the Exploratory Structural Equation Modeling Framework
No full textWording effects, the systematic method variance arising from the inconsistent responding to positively and negatively worded items of the same construct, are pervasive in the behavioral and health sciences. Although several factor modeling strategies have been proposed to mitigate their adverse effects, there is limited systematic research assessing their performance with exploratory structural equation models (ESEM). The present study evaluated the impact of different types of response bias related to wording effects (random and straight-line carelessness, acquiescence, item difficulty, and mixed) on ESEM models incorporating two popular method modeling strategies, the correlated traits-correlated methods minus one (CTC[M-1]) model and random intercept item factor analysis (RIIFA), as well as the “do nothing” approach. Five variables were manipulated using Monte Carlo methods: the type and magnitude of response bias, factor loadings, factor correlations, and sample size. Overall, the results showed that ignoring wording effects leads to poor model fit and serious distortions of the ESEM estimates. The RIIFA approach generally performed best at countering these adverse impacts and recovering unbiased factor structures, whereas the CTC(M-1) models struggled when biases affected both positively and negatively worded items. A straightforward guide is offered to applied researchers who wish to use ESEM with mixed-worded scales
What do we measure when we measure pubertal development? Problems and solutions in the conceptual basis of pubertal measures
No full textPubertal development influences adolescent health and behavior, but measuring puberty is challenging. Difficulties stem not only from practical issues, which have been the extensively discussed in the literature, but also from the very complexity of the theoretical basis of pubertal development from a physiological point of view. Here we discuss conceptual issues related to measuring and interpreting pubertal effects by first analyzing a widely used measure – age at menarche – as a case study. We use age at menarche as a starting point not because it is an ideal measure, but precisely because it is not, being well known for its limitations. However, our analysis shows that the most significant limitations this measure also apply to all currently used pubertal measures and arise from the fundamentally multifactorial physiological nature of pubertal development. Puberty is not a unitary phenomenon, cannot be measured directly nor be defined by individual makers, and its different markers are often discordant. We argue that a possible compromise between practical issues (building feasible/reliable measures) and theoretical ones (building conceptually sound and interpretable measures) is to treat puberty as the variance shared among multiple pubertal measures using suitable statistical methods
Count regression models for keyness analysis
No full textA wide variety of measures have been used in previous work to assess the keyness of items in a particular domain of language use. The present paper explores an approach to keyword analysis based on regression modeling. Specifically, we use a form of negative binomial regression, which offers a number of advantages compared to existing techniques for identifying typical items in a target corpus. Thus, it is responsive to the multidimensional nature of keyness and can address multiple aspects of typicalness simultaneously, using a single statistical model. Further, metrics of interest can be enriched with confidence intervals, which allows us to isolate descriptive and inferential indicators of keyness. Finally, all quantities are based on a text-level analysis, which accounts for the fact that the target and reference corpus consist of text files and adjusts uncertainty estimates accordingly. As an illustrative case study, we rely on COCA to identify key verbs in academic writing and demonstrate how negative binomial regression may be used to this end. Our checks on the coverage rate of the 95% confidence intervals indicate that this model seems to be adequate for purposes of statistical inference. Due consideration will also be given to the limitations of this procedure, and we conclude by outlining the kinds of keyness analyses for which count regression models may be a worthwhile approach. The online supplementary material for this paper provides data and R code for the implementation of keyness regression
We need to explain subjective experience, but its explanation may not be mechanistic
No full textModels of consciousness need to explain both the objective correlates of conscious experience as well as its subjective structure. However, such an explanation would not need to entail a reduction exclusively in terms of physical or neural systems. A model that intends to avoid such reduction is integrated information theory (IIT). In this article, we discuss the explanatory rationale of IIT, its potential inconsistencies and its consequences for the neuroscience of consciousness more broadly. In particular, we identify ambiguities regarding the directionality of the explanation, i.e., important tensions between IIT's purported ontological and epistemological primacy of experience, and its explanatory aim of accounting for consciousness in physical, operational terms. Across the text, we propose several ways to avoid these issues and eventually complement, enhance or replace the model. The main goal is to motivate clarification among IIT-proponents and inform IIT-opponents on accurate points of contention, without thereby misrepresenting the model. In our final section, we introduce alternative explanatory paths: mathematical, processual, and autonomy-based types of explanations. These novel and sound explanatory strategies may better inspire the next generation of models of consciousness