Simulation of Human Behavior

How even the greatest scientists can fall for cognitive bias

New Scientist

A HUNDRED years ago, scientists were sure of many truths. The greatest experts were certain that the universe had always existed and was always the size it is now. Most biologists were sure that proteins, not DNA, were responsible for heredity. Biochemists believed that, outside the nucleus, the interior of a human cell contained little more than busy enzymes – a "biochemical bog" that carried out all the reactions that are essential to life. These sureties were all wrong.

Scientists create a model human embryo with its very own heartbeat

Daily Mail - Science & tech

A model human embryo with a heartbeat and traces of blood has been created by scientists in a move that could offer insights into the first weeks of life. The synthetic structure was created from human stem cells without the need for eggs, sperm or fertilisation. It replicates some of the cells and structures that would usually appear in the third and fourth weeks of pregnancy, but was designed to never have the ability to develop into a foetus. Despite the heartbeat, the structure does not have the tissues that go on to form the placenta and yolk sac in a natural embryo. 'I'd like to emphasise that these are neither embryos nor are we trying to make embryos,' said Dr Jitesh Neupane, from the University of Cambridge's Gurdon Institute.

Deep Learning for Predicting Human Strategic Behavior

Neural Information Processing Systems

Predicting the behavior of human participants in strategic settings is an important problem in many domains. Most existing work either assumes that participants are perfectly rational, or attempts to directly model each participant's cognitive processes based on insights from cognitive psychology and experimental economics. In this work, we present an alternative, a deep learning approach that automatically performs cognitive modeling without relying on such expert knowledge. We introduce a novel architecture that allows a single network to generalize across different input and output dimensions by using matrix units rather than scalar units, and show that its performance significantly outperforms that of the previous state of the art, which relies on expert-constructed features.

Improving Natural Language Processing Tasks with Human Gaze-Guided Neural Attention University of Stuttgart, Institute for Visualization and Interactive Systems (VIS), Germany

Neural Information Processing Systems

A lack of corpora has so far limited advances in integrating human gaze data as a supervisory signal in neural attention mechanisms for natural language processing (NLP). We propose a novel hybrid text saliency model (TSM) that, for the first time, combines a cognitive model of reading with explicit human gaze supervision in a single machine learning framework. On four different corpora we demonstrate that our hybrid TSM duration predictions are highly correlated with human gaze ground truth. We further propose a novel joint modeling approach to integrate TSM predictions into the attention layer of a network designed for a specific upstream NLP task without the need for any task-specific human gaze data. We demonstrate that our joint model outperforms the state of the art in paraphrase generation on the Quora Question Pairs corpus by more than 10% in BLEU-4 and achieves state of the art performance for sentence compression on the challenging Google Sentence Compression corpus. As such, our work introduces a practical approach for bridging between data-driven and cognitive models and demonstrates a new way to integrate human gaze-guided neural attention into NLP tasks.

Capturing Failures of Large Language Models via Human Cognitive Biases Jacob Steinhardt UC Berkeley

Neural Information Processing Systems

Large language models generate complex, open-ended outputs: instead of outputting a class label they write summaries, generate dialogue, or produce working code. In order to asses the reliability of these open-ended generation systems, we aim to identify qualitative categories of erroneous behavior, beyond identifying individual errors. To hypothesize and test for such qualitative errors, we draw inspiration from human cognitive biases--systematic patterns of deviation from rational judgement. Specifically, we use cognitive biases as motivation to (i) generate hypotheses for problems that models may have, and (ii) develop experiments that elicit these problems. Using code generation as a case study, we find that OpenAI's Codex errs predictably based on how the input prompt is framed, adjusts outputs towards anchors, and is biased towards outputs that mimic frequent training examples. We then use our framework to elicit high-impact errors such as incorrectly deleting files. Our results indicate that experimental methodology from cognitive science can help characterize how machine learning systems behave.

Classification in Non-Metric Spaces

Neural Information Processing Systems

A key question in vision is how to represent our knowledge of previously encountered objects to classify new ones. The answer depends on how we determine the similarity of two objects. Similarity tells us how relevant each previously seen object is in determining the category to which a new object belongs. Complex notions of similar(cid:173) ity appear necessary for cognitive models and applications, while simple notions of similarity form a tractable basis for current computational ap(cid:173) proaches to classification. We explore the nature of this dichotomy and why it calls for new approaches to well-studied problems in learning. We begin this process by demonstrating new computational methods for supervised learning that can handle complex notions of similarity.

Constructing Distributed Representations Using Additive Clustering

Neural Information Processing Systems

Many cognitive models posit mental representations based on discrete substructures. Even connectionist models whose processing involves manipulation of real-valued activations typically represent objects as patterns of 0s and 1s across a set of units (Noelle, Cottrell, and Wilms, 1997). Often, individual units are taken to represent specific features of the objects and two representations will share features to the degree to which the two objects are similar. While this arrangement is intuitively appealing, it can be difficult to construct the features to be used in such a model. Using random feature assignments clouds the relationship between the model and the objects it is intended to represent, diminishing the model's value. As Clouse and Cottrell (1996) point out, hand-crafted representations are tedious to construct and it can be difficult to precisely justify (or even articulate) the principles that guided their design. These difficulties effectively limit the number of objects that can be encoded, constraining modeling efforts to small examples. In this paper, we investigate methods for automatically synthesizing feature-based representations directly from the pairwise object similarities that the model is intended to respect.

Searching for Character Models

Neural Information Processing Systems

We introduce a method to automatically improve character models for a handwritten script without the use of transcriptions and using a minimum of document specific training data. We show that we can use searches for the words in a dictionary to identify portions of the document whose transcriptions are unambiguous. Using templates extracted from those regions, we retrain our character prediction model to drastically improve our search retrieval performance for words in the document.

Inducing Metric Violations in Human Similarity Judgements

Neural Information Processing Systems

Attempting to model human categorization and similarity judgements is both a very interesting but also an exceedingly difficult challenge. Some of the difficulty arises because of conflicting evidence whether human categorization and similarity judgements should or should not be modelled as to operate on a mental representation that is essentially metric. Intuitively, this has a strong appeal as it would allow (dis)similarity to be represented geometrically as distance in some internal space. Here we show how a single stimulus, carefully constructed in a psychophysical experiment, introduces l2 violations in what used to be an internal similarity space that could be adequately modelled as Euclidean. We term this one influential data point a conflictual judgement.

Multi-scale Hyper-time Hardware Emulation of Human Motor Nervous System Based on Spiking Neurons using FPGA

Neural Information Processing Systems

Our central goal is to quantify the long-term progression of pediatric neurological diseases, such as a typical 10-15 years progression of child dystonia. To this purpose, quantitative models are convincing only if they can provide multi-scale details ranging from neuron spikes to limb biomechanics. The models also need to be evaluated in hyper-time, i.e. significantly faster than real-time, for producing useful predictions. We designed a platform with digital VLSI hardware for multi-scale hyper-time emulations of human motor nervous systems. The platform is constructed on a scalable, distributed array of Field Programmable Gate Array (FPGA) devices.