Studies with impossible languages falsify LMs as models of human language Jeffrey S. Bowers, School of Psychology and Neuroscience, University of Bristol Jeff Mitchell, School of Engineering and Informatics, University of Sussex Commentary on Futrell, R., & Mahowald, K. (in press). How linguistics learned to stop worrying and love the language models. Abstract According to Futrell and Mahowald (F&M), both infants and language models (LMs) find attested languages easier to learn than "impossible languages" that have unnatural structures. We review the literature and show that LMs often learn attested and many impossible languages equally well. Difficult to learn impossible languages are simply more complex (or random).

Argumentation frameworks (AFs) provide formal approaches for legal reasoning, but identifying sources of ambiguity and explaining argument acceptance remains challenging for non-experts. We present AF-XRAY, an open-source toolkit for exploring, analyzing, and visualizing abstract AFs in legal reasoning. AF-XRAY introduces: (i) layered visualizations based on game-theoretic argument length revealing well-founded derivation structures; (ii) classification of attack edges by semantic roles (primary, secondary, blunders); (iii) overlay visualizations of alternative 2-valued solutions on ambiguous 3-valued grounded semantics; and (iv) identification of critical attack sets whose suspension resolves undecided arguments. Through systematic generation of critical attack sets, AF-XRAY transforms ambiguous scenarios into grounded solutions, enabling users to pinpoint specific causes of ambiguity and explore alternative resolutions. We use real-world legal cases (e.g., Wild Animals as modeled by Bench-Capon) to show that our tool supports teleological legal reasoning by revealing how different assumptions lead to different justified conclusions.

While convolutional neural networks (CNNs) have come to match and exceed human performance in many settings, the tasks these models optimize for are largely constrained to the level of individual objects, such as classification and captioning. Humans remain vastly superior to CNNs in visual tasks involving relations, including the ability to identify two objects as `same' or `different'. A number of studies have shown that while CNNs can be coaxed into learning the same-different relation in some settings, they tend to generalize poorly to other instances of this relation. In this work we show that the same CNN architectures that fail to generalize the same-different relation with conventional training are able to succeed when trained via meta-learning, which explicitly encourages abstraction and generalization across tasks.

Federal officials are investigating how a Tesla Cybertruck crash killed three people in Northern California last month. A National Highway Traffic Safety Administration spokesperson confirmed that the agency is aware of the crash and is gathering information from law enforcement and the manufacturer. The NHTSA is the agency in charge of reviewing incidents involving automated driving technology. The Cybertruck comes with Tesla's Autopilot driving feature, which includes driver-assistance technology, and the Full Self-Driving system is optional. It is unclear if the driver was using the Autopilot feature at the time of the accident.

Astronauts aboard the International Space Station (ISS) had slower memory, attention and processing speed after six months, raising concerns about the impact of cognitive impairment on future space missions to Mars. The extreme environment of space, with reduced gravity, harsh radiation and the lack of regular sunrises and sunsets, can have dramatic effects on astronaut health, from muscle loss to an increased risk of heart disease. However, the cognitive effects of long-term space travel are less well documented. Inside NASA's ambitious plan to bring the ISS crashing back to Earth Now, Sheena Dev at NASA's Johnson Space Center in Houston, Texas, and her colleagues have looked at the cognitive performance of 25 astronauts during their time on the ISS. The team ran the astronauts through 10 tests, some of which were done on Earth, once before and twice after the mission, while others were done on the ISS, both early and later in the mission.

Neither the hype exemplified in some exaggerated claims about deep neural networks (DNNs), nor the gloom expressed by Bowers et al. do DNNs as models in vision science justice: DNNs rapidly evolve, and today's limitations are often tomorrow's successes. In addition, providing explanations as well as prediction and image-computability are model desiderata; one should not be favoured at the expense of the other. We agree with Bowers et al. (2022) that some of the quoted statements at the beginning of their target article about DNNs as "best models" are exaggerated--perhaps some of them bordering on scientific hype (Intemann, 2020). However, only the authors of such exaggerated statements are to blame, not DNNs: Instead of blaming DNNs, perhaps Bowers et al. should have engaged in a critical discussion of the increasingly widespread practice of rewarding impact and boldness over carefulness and modesty that allows hyperbole to flourish in science. This is unfortunate as the target article does mention a number of valid issues with DNNs in vision science and raises a number of valid concerns. For example, we fully agree that human vision is much more than recognising photographs of objects in scenes; we also fully agree there are still a number of important behavioural differences between DNNs and humans even in terms of core object recognition (DiCarlo et al., 2012), i.e. even when recognising photographs of objects in scenes, such as DNNs' adversarial susceptibility (section 4.1.1)

Visual illusions remind us that we are not passive decoders of reality but active interpreters. Our eyes capture information from the environment, but our brain can play tricks on us. Perception doesn't always match reality. Scientists have used illusions for decades to explore the psychological and cognitive processes that underlie human visual perception. More recently, evidence is emerging that suggests many animals, like us, can perceive and create a range of visual illusions.

Apple has been incredibly secretive about its efforts to develop a self-driving car, but according to Bloomberg, its latest move is hiring a key personnel from its toughest competition. The tech giant has reportedly hired Christopher "CJ" Moore, who's been the director for Tesla's Autopilot Software since 2019 and who's been with the company since 2014. While Moore has yet to update his LinkedIn page, Bloomberg says he will also work on software at Apple and will report to Stuart Bowers. Like Moore, Bowers worked as Tesla's head of Autopilot unit until he left in 2019. Back in May, Moore was one of the employees who told the California DMV that Elon Musk exaggerated the automaker's full self-driving timeline.

Humans can identify objects following various spatial transformations such as scale and viewpoint. This extends to novel objects, after a single presentation at a single pose, sometimes referred to as online invariance. CNNs have been proposed as a compelling model of human vision, but their ability to identify objects across transformations is typically tested on held-out samples of trained categories after extensive data augmentation. This paper assesses whether standard CNNs can support human-like online invariance by training models to recognize images of synthetic 3D objects that undergo several transformations: rotation, scaling, translation, brightness, contrast, and viewpoint. Through the analysis of models' internal representations, we show that standard supervised CNNs trained on transformed objects can acquire strong invariances on novel classes even when trained with as few as 50 objects taken from 10 classes. This extended to a different dataset of photographs of real objects. We also show that these invariances can be acquired in a self-supervised way, through solving the same/different task. We suggest that this latter approach may be similar to how humans acquire invariances.

Male bowerbirds in Australia use a technique called forced perspective to make themselves look bigger to potential mates who visit their carefully constructed bowers. Visual illusions remind us that we are not passive decoders of reality but active interpreters. Our eyes capture information from the environment, but our brain can play tricks on us. Perception doesn't always match reality. Scientists have used illusions for decades to explore the psychological and cognitive processes that underlie human visual perception.