Large Language Models (LLMs) offer a framework for understanding language processing in the human brain. Unlike traditional models, LLMs represent words and context through layered numerical embeddings. Here, we demonstrate that LLMs' layer hierarchy aligns with the temporal dynamics of language comprehension in the brain. Using electrocorticography (ECoG) data from participants listening to a 30-minute narrative, we show that deeper LLM layers correspond to later brain activity, particularly in Broca's area and other language-related regions. We extract contextual embeddings from GPT-2 XL and Llama-2 and use linear models to predict neural responses across time. Our results reveal a strong correlation between model depth and the brain's temporal receptive window during comprehension. We also compare LLM-based predictions with symbolic approaches, highlighting the advantages of deep learning models in capturing brain dynamics. We release our aligned neural and linguistic dataset as a public benchmark to test competing theories of language processing.

Contextual embeddings, derived from deep language models (DLMs), provide a continuous vectorial representation of language. This embedding space differs fundamentally from the symbolic representations posited by traditional psycholinguistics. We hypothesize that language areas in the human brain, similar to DLMs, rely on a continuous embedding space to represent language. To test this hypothesis, we densely record the neural activity patterns in the inferior frontal gyrus (IFG) of three participants using dense intracranial arrays while they listened to a 30-minute podcast. From these fine-grained spatiotemporal neural recordings, we derive a continuous vectorial representation for each word (i.e., a brain embedding) in each patient. Using stringent zero-shot mapping we demonstrate that brain embeddings in the IFG and the DLM contextual embedding space have common geometric patterns. The common geometric patterns allow us to predict the brain embedding in IFG of a given left-out word based solely on its geometrical relationship to other non-overlapping words in the podcast. Furthermore, we show that contextual embeddings capture the geometry of IFG embeddings better than static word embeddings. The continuous brain embedding space exposes a vector-based neural code for natural language processing in the human brain. Here, using neural activity patterns in the inferior frontal gyrus and large language modeling embeddings, the authors provide evidence for a common neural code for language processing.

Deep Language Models (DLMs) provide a novel computational paradigm for understanding the mechanisms of natural language processing in the human brain. Unlike traditional psycholinguistic models, DLMs use layered sequences of continuous numerical vectors to represent words and context, allowing a plethora of emerging applications such as human-like text generation. In this paper we show evidence that the layered hierarchy of DLMs may be used to model the temporal dynamics of language comprehension in the brain by demonstrating a strong correlation between DLM layer depth and the time at which layers are most predictive of the human brain. Our ability to temporally resolve individual layers benefits from our use of electrocorticography (ECoG) data, which has a much higher temporal resolution than noninvasive methods like fMRI. Using ECoG, we record neural activity from participants listening to a 30-minute narrative while also feeding the same narrative to a high-performing DLM (GPT2-XL). We then extract contextual embeddings from the different layers of the DLM and use linear encoding models to predict neural activity. We first focus on the Inferior Frontal Gyrus (IFG, or Broca's area) and then extend our model to track the increasing temporal receptive window along the linguistic processing hierarchy from auditory to syntactic and semantic areas. Our results reveal a connection between human language processing and DLMs, with the DLM's layer-by-layer accumulation of contextual information mirroring the timing of neural activity in high-order language areas.

AlexNet is a pretrained convolutional neural network (CNN) that has been trained on approximately 1.2 million images from the ImageNet Dataset ( http://image-net.org/index ). The model has 25 layers and can classify images into 1000 object categories. It can be used for all sorts of object classification. However, if an object was not in the training set, it won't be able to identify the object. If a banana was in the training set, you could expect the CNN to correctly identify a new picture of a banana. But if you gave it a picture of a plantain, and plantain was NOT in the CNN, then it might not find a match or, more likely, it might incorrectly classify it like a banana.

Earth sensors were some of the earliest spacecraft sensors. Many early satellites were for communications or Earth observations, so the most important thing was to point the satellite at the Earth, that is, to null the roll and pitch errors. In many cases, the yaw error was not important or could be controlled through the momentum of the satellite. Earth sensors were either static, without moving parts, or scanning. Scanning was done by rotating the satellite, rotating a mirror with a motor, or rotating a mirror with a torsion bar. Most sensors operated in the infrared wavelengths because the radiation at those wavelengths was more uniform than that in the visible bands and is unaffected by eclipses. Many CubeSats use static Earth sensors because they are simple, compact, and inexpensive. A wide variety of models are available.

Contextual embeddings, derived from deep language models (DLMs), provide a continuous vectorial representation of language. This embedding space differs fundamentally from the symbolic representations posited by traditional psycholinguistics. Do language areas in the human brain, similar to DLMs, rely on a continuous embedding space to represent language? To test this hypothesis, we densely recorded the neural activity in the Inferior Frontal Gyrus (IFG, also known as Broca’s area) of three participants using dense intracranial arrays while they listened to a 30-minute podcast. From these fine-grained spatiotemporal neural recordings, we derived for each patient a continuous vectorial representation for each word (i.e., a brain embedding). Using stringent, zero-shot mapping, we demonstrated that brain embeddings in the IFG and the DLM contextual embedding space have strikingly similar geometry. This shared geometry allows us to precisely triangulate the position of unseen words in both the brain embedding space (zero-shot encoding) and the DLM contextual embedding space (zero-shot decoding). The continuous brain embedding space provides an alternative computational framework for how natural language is represented in cortical language areas.

Generative machine learning (ML) models are a class of models that allow you to create new data by modeling the data generating distribution. For example, a generative model trained on images of human faces would learn what features constitute a realistic human face and how to combine them to generate novel human face images. For a fun demonstration of the power of ML-based human face generation, check out [44].

Netflix, Hulu, and Amazon Prime all attempt to help you pick movies. In this chapter, we will create a database of movies, with fictional ratings. We will then create a set of viewers. We will then try to predict if a viewer would choose to watch a particular movie. We will use deep learning with MATLAB’s pattern recognition network, patternnet. You will see that we can achieve accuracies of up to 100% over our small set of movies. Guessing what a customer would like to buy is something that all manufacturers and retailers want to do as it lets them focus their efforts on products that are of the greatest interest to their customers. As we show in this chapter, deep learning can be a valuable tool.

Tokamaks are fusion machines that are under development to produce baseload power. Baseload power is the power that is produced 24/7 and provides the base for powering the electric grid. The International Thermonuclear Experimental Reactor (ITER) is an international project that will produce net power from a Tokamak. Net power means the Tokamak produces more energy than it consumes. Consumption includes heating the plasma, controlling it, and powering all the auxiliary systems needed to maintain the plasma. It will allow researchers to study the physics of the Tokamak which will hopefully lead the way toward operational machines.

Deep learning is a subset of machine learning which is itself a subset of artificial intelligence and statistics. Artificial intelligence research began shortly after World War II [24]. Early work was based on the knowledge of the structure of the brain, propositional logic, and Turing's theory of computation. Warren McCulloch and Walter Pitts created a mathematical formulation for neural networks based on threshold logic. This allowed neural network research to split into two approaches: one centered on biological processes in the brain and the other on the application of neural networks to artificial intelligence. It was demonstrated that any function could be implemented through a set of such neurons and that a neural net could learn. In 1948, Norbert Wiener's book, Cybernetics, was published which described concepts in control, communications, and statistical signal processing. The next major step in neural networks was Donald Hebb's book in 1949, The Organization of Behavior, connecting connectivity with learning in the brain. His book became a source of learning and adaptive systems. Marvin Minsky and Dean Edmonds built the first neural computer at Harvard in 1950.

A pirouette is a familiar step in ballet. There are many types of pirouettes. We will focus on an en dehors (outside) pirouette from the fourth position. The dancer pliés (does a deep knee bend) and then straightens their legs, producing both an upward force to get on the tip of their pointe shoe and torque to turn about their axis of revolution.

The Direct Fusion Drive (DFD), a compact, anuetronic fusion engine, will enable more challenging exploration missions in the solar system. The engine proposed here uses a deuterium–helium-3 reaction to produce fusion energy by employing a novel field-reversed configuration (FRC) for magnetic confinement. The FRC has a simple linear solenoid coil geometry yet generates higher plasma pressure, hence higher fusion power density, for a given magnetic field strength than other magnetic-confinement plasma devices. Waste heat generated from the plasma׳s Bremsstrahlung and synchrotron radiation is recycled to maintain the fusion temperature. The charged reaction products, augmented by additional propellant, are exhausted through a magnetic nozzle. A 1MW DFD is presented in the context of a mission to deploy the James Webb Space Telescope (6200kg) from GPS orbit to a Sun–Earth L2 halo orbit in 37 days using just 353kg of propellant and about half a kilogram of 3He. The engine is designed to produce 40N of thrust with an exhaust velocity of 56.5km/s and has a specific power of 0.18kW/kg.