Reviewer Recognition
Related Articles
Testing the reliability of an AI-based large language model to extract ecological information from the scientific literature
Artificial intelligence-based large language models (LLMs) have the potential to substantially improve the efficiency and scale of ecological research, but their propensity for delivering incorrect information raises significant concern about their usefulness in their current state. Here, we formally test how quickly and accurately an LLM performs in comparison to a human reviewer when tasked with extracting various types of ecological data from the scientific literature. We found the LLM was able to extract relevant data over 50 times faster than the reviewer and had very high accuracy (>90%) in extracting discrete and categorical data, but it performed poorly when extracting certain quantitative data. Our case study shows that LLMs offer great potential for generating large ecological databases at unprecedented speed and scale, but additional quality assurance steps are required to ensure data integrity.
Interracial contact shapes racial bias in the learning of person-knowledge
During impression formation, perceptual cues facilitate social categorization while person-knowledge can promote individuation and enhance person memory. Although there is extensive literature on the cross-race recognition deficit, observed when racial ingroup faces are recognized more than outgroup faces, it is unclear whether a similar deficit exists when recalling individuating information about outgroup members. To better understand how perceived race can bias person memory, the present study examined how self-identified White perceivers’ interracial contact impacts learning of perceptual cues and person-knowledge about perceived Black and White others over five sessions of training. While person-knowledge facilitated face recognition accuracy for low-contact perceivers, face recognition accuracy did not differ for high-contact perceivers based on person-knowledge availability. The results indicate a bias towards better recall of ingroup person knowledge, which decreased for high-contact perceivers across the five-day training but simultaneously increased for low-contact perceivers. Overall, the elimination of racial bias in recall of person-knowledge among high-contact perceivers amid a persistent cross-race deficit in face recognition suggests that contact may have a greater impact on the recall of person-knowledge than on face recognition.
Value of clinical review for AI-guided deep vein thrombosis diagnosis with ultrasound imaging by non-expert operators
Deep vein thrombosis (DVT) carries high morbidity, mortality, and costs globally. Point of care ultrasound (POCUS) image acquisition by non-ultrasound-trained providers, supported by an AI-based guidance and remote image review system, is believed to improve the timeliness and cost-effectiveness of diagnosis. We examine a database of 381 patients with suspected DVT who underwent an AI-guided ultrasound scan by a non-ultrasound-trained nurse and an expert sonographer-performed standard compression ultrasound scan. Each AI-guided scan was reviewed remotely by blinded radiologists or blinded independent POCUS-certified American Emergency Medicine (EM) physicians. Remote reviewer and standard scan diagnoses were compared. The primary endpoint is AI-guidance system sensitivity with clinician review; secondary endpoints include specificity, positive predictive value, negative predictive value, image quality, inter-observer image quality, and vein compressibility agreement. Data was analysed through the bootstrapping method, bootstrapping with a second reader for each scan, and a majority voting system. Eighty percent (n = 304) of scans were of sufficient diagnostic quality. Radiologist reviewer sensitivity ranged from 90%–95%, specificity from 74%–84%, NPV from 98%–99%, PPV from 30%–42%, and potential expert-led ultrasound scans avoided from 39%–50%. Inter-observer agreement for image quality was 0.15 and for compressibility 0.61. EM reviewer sensitivity ranged from 95%–98%, specificity from 97%–100%, NPV was 99%, PPV from 81%–100%, and potential expert-led ultrasound scans avoided from 29%–38%. Inter-observer agreement for image quality was 0.59 and for compressibility 0.67. Diagnosing lower extremity DVT through AI-guided image acquisition with clinician review is feasible. Performance is influenced by reviewer expertise. We find potential positive impacts on health economics, including safely avoiding expert-led ultrasound scans.
Predictive learning as the basis of the testing effect
A prominent learning phenomenon is the testing effect, meaning that testing enhances retention more than studying. Emergent frameworks propose fundamental (Hebbian and predictive) learning principles as its basis. Predictive learning posits that learning occurs based on the contrast (error) between a prediction and the feedback on that prediction (prediction error). Here, we propose that in testing (but not studying) scenarios, participants predict potential answers, and its contrast with the subsequent feedback yields a prediction error, which facilitates testing-based learning. To investigate this, we developed an associative memory network incorporating Hebbian and/or predictive learning, together with an experimental design where human participants studied or tested English-Swahili word pairs followed by recognition. Three behavioral experiments (N = 80, 81, 62) showed robust testing effects when feedback was provided. Model fitting (of 10 different models) suggested that only models incorporating predictive learning can account for the breadth of data associated with the testing effect. Our data and model suggest that predictive learning underlies the testing effect.
Configural processing as an optimized strategy for robust object recognition in neural networks
Configural processing, the perception of spatial relationships among an object’s components, is crucial for object recognition, yet its teleology and underlying mechanisms remain unclear. We hypothesize that configural processing drives robust recognition under varying conditions. Using identification tasks with composite letter stimuli, we compare neural network models trained with either configural or local cues. We find that configural cues support robust generalization across geometric transformations (e.g., rotation, scaling) and novel feature sets. When both cues are available, configural cues dominate local features. Layerwise analysis reveals that sensitivity to configural cues emerges later in processing, likely enhancing robustness to pixel-level transformations. Notably, this occurs in a purely feedforward manner without recurrent computations. These findings with letter stimuli successfully extend to naturalistic face images. Our results demonstrate that configural processing emerges in a naíve network based on task contingencies, and is beneficial for robust object processing under varying viewing conditions.
Responses