A prerequisite for social coordination is bidirectional communication between teammates, each playing two roles simultaneously: as receptive listeners and expressive speakers. For robots working with humans in complex situations with multiple goals that differ in importance, failure to fulfill the expectation of either role could undermine group performance due to misalignment of values between humans and robots. Specifically, a robot needs to serve as an effective listener to infer human users’ intents from instructions and feedback and as an expressive speaker to explain its decision processes to users. Here, we investigate how to foster effective bidirectional human-robot communications in the context of value alignment—collaborative robots and users form an aligned understanding of the importance of possible task goals. We propose an explainable artificial intelligence (XAI)
system in which a group of robots predicts users’ values by taking in situ feedback into consideration while communicating their decision processes to users through explanations. To learn from human feedback, our XAI system integrates a cooperative communication model for inferring human values associated with multiple desirable goals. To be interpretable to humans, the system simulates human mental dynamics and predicts optimal explanations using graphical models. We conducted psychological experiments to examine the core components of the proposed computational framework. Our results show that real-time human-robot mutual understanding in complex cooperative tasks is achievable with a learning model based on bidirectional communication. We believe that this interaction framework can shed light on bidirectional value alignment in communicative XAI systems and, more broadly, in future human-machine teaming systems.
more »
« less
Learning to Mediate Disparities Towards Pragmatic Communication
Human communication is a collaborative process. Speakers, on top of conveying their own intent, adjust the content and language expressions by taking the listeners into account, including their knowledge background, personalities, and physical capabilities. Towards building AI agents with similar abilities in language communication, we propose Pragmatic Rational Speaker (PRS), a framework extending Rational Speech Act (RSA). The PRS attempts to learn the speaker-listener disparity and adjust the speech accordingly, by adding a light-weighted disparity adjustment layer into working memory on top of speaker’s long-term memory system. By fixing the long-term memory, the PRS only needs to update its working memory to learn and adapt to different types of listeners. To validate our framework, we create a dataset that simulates different types of speaker-listener disparities in the context of referential games. Our empirical results demonstrate that the PRS is able to shift its output towards the language that listeners are able to understand, significantly improve the collaborative task outcome.
more »
« less
- Award ID(s):
- 1949634
- NSF-PAR ID:
- 10321126
- Date Published:
- Journal Name:
- Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (ACL)
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
more » « less
Abstract For nearly 25 years, researchers have recognized the rich and numerous facets of native perception of non‐native speech, driving a large, and growing, body of work that has shed light on how native listeners understand non‐native speech. The bulk of this work, however, has focused on the talker. That is, most researchers have asked what perception of non‐native speech tells us about the non‐native speaker, or when interacting with non‐native speakers more generally. It is clear that listeners perceive speech not only in terms of the acoustic signal, but also with their own experience and biases driving their perception. It is also clear that native listeners can improve their perception of non‐native speech for both familiar and unfamiliar accents. Therefore, it is imperative that research in non‐native communication also consider an active role for the listener. To truly understand communication between native and non‐native speakers, it is critically important to understand both the properties of non‐native speech and how this speech is perceived. In the present review, we describe non‐native speech and then review previous research, examining the methodological shift from using native listeners as tools to understand properties of non‐native speech to understanding listeners as partners in conversation. We discuss how current models not only limit our understanding of non‐native speech, but also limit what types of questions researchers set out to answer. We demonstrate that while non‐native speakers capable of shifting their productions to be better understood by listeners, native listeners are also capable of shifting their perception to more accurately perceive non‐native speech. We conclude by setting forth a series of recommendations for future research, emphasizing the contributions of native listeners and non‐native speakers as equally important for communicative success.
-
more » « less
Abstract Despite the
lack of invariance problem (the many‐to‐many mapping between acoustics and percepts), human listeners experience phonetic constancy and typically perceive what a speaker intends. Most models of human speech recognition (HSR) have side‐stepped this problem, working with abstract, idealized inputs and deferring the challenge of working with real speech. In contrast, carefully engineered deep learning networks allow robust, real‐world automatic speech recognition (ASR). However, the complexities of deep learning architectures and training regimens make it difficult to use them to provide direct insights into mechanisms that may support HSR. In this brief article, we report preliminary results from a two‐layer network that borrows one element from ASR,long short‐term memory nodes, which provide dynamic memory for a range of temporal spans. This allows the model to learn to map real speech from multiple talkers to semantic targets with high accuracy, with human‐like timecourse of lexical access and phonological competition. Internal representations emerge that resemble phonetically organized responses in human superior temporal gyrus, suggesting that the model develops a distributed phonological code despite no explicit training on phonetic or phonemic targets. The ability to work with real speech is a major advance for cognitive models of HSR. -
For much of its history, categorical perception was treated as a foundational theory of speech perception, which suggested that quasi-discrete categorization was a goal of speech perception. This had a profound impact on bilingualism research which adopted similar tasks to use as measures of nativeness or native-like processing, implicitly assuming that any deviation from discreteness was a deficit. This is particularly problematic for listeners like heritage speakers whose language proficiency, both in their heritage language and their majority language, is questioned. However, we now know that in the monolingual listener, speech perception is gradient and listeners use this gradiency to adjust subphonetic details, recover from ambiguity, and aid learning and adaptation. This calls for new theoretical and methodological approaches to bilingualism. We present the Visual Analogue Scaling task which avoids the discrete and binary assumptions of categorical perception and can capture gradiency more precisely than other measures. Our goal is to provide bilingualism researchers new conceptual and empirical tools that can help examine speech categorization in different bilingual communities without the necessity of forcing their speech categorization into discrete units and without assuming a deficit model.more » « less
-
more » « less
Abstract Prior work in speech processing indicates that listening tasks with multiple speakers (as opposed to a single speaker) result in slower and less accurate processing. Notably, the trial-to-trial cognitive demands of switching between speakers or switching between accents have yet to be examined. We used pupillometry, a physiological index of cognitive load, to examine the demands of processing first (L1) and second (L2) language-accented speech when listening to sentences produced by the same speaker consecutively (no switch), a novel speaker of the same accent (within-accent switch), and a novel speaker with a different accent (across-accent switch). Inspired by research on sequential adjustments in cognitive control, we aimed to identify the cognitive demands of accommodating a novel speaker and accent by examining the trial-to-trial changes in pupil dilation during speech processing. Our results indicate that switching between speakers was more cognitively demanding than listening to the same speaker consecutively. Additionally, switching to a novel speaker with a different accent was more cognitively demanding than switching between speakers of the same accent. However, there was an asymmetry for across-accent switches, such that switching from an L1 to an L2 accent was more demanding than vice versa. Findings from the present study align with work examining multi-talker processing costs, and provide novel evidence that listeners dynamically adjust cognitive processing to accommodate speaker and accent variability. We discuss these novel findings in the context of an active control model and auditory streaming framework of speech processing.
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- The DOI is not currently available.
