Summary: Researchers have shown that people can infer water temperature simply by listening to the sound of it being poured. Using machine learning and computational auditory analysis, the study demonstrates how listeners unconsciously use acoustic cues to perceive thermal properties, revealing a subtle cross-modal sensory ability with potential applications in neuroscience and sensory technology.
This research highlights an implicit perceptual skill that develops through life experience and underscores the value of combining human psychophysics with artificial intelligence to clarify complex multisensory mappings. The findings may inform future work in auditory neuroscience, sensory augmentation, and human–computer interaction.
Key Facts:
- New Skill: Humans can distinguish the temperature of water by the sound it makes when poured.
- Machine Learning: A deep neural network and a classifier reliably identified hot versus cold water from audio recordings.
- Implicit Learning: The ability appears to be learned implicitly through lifelong exposure to acoustic differences tied to temperature.
Source: Reichman University
Researchers at the Ivcher Institute for Brain, Cognition, and Technology (BCT Institute) at Reichman University (IDC Herzliya) investigated a little-known perceptual capacity: whether humans can perceive thermal properties through audition, and how this cross-modal ability is represented.
The team framed the work within multisensory integration theory—the way the brain combines inputs from different senses to form coherent perceptions—and focused specifically on thermal perception conveyed by auditory cues. They recorded pouring sounds of water at various temperatures and analyzed those recordings both with human listeners and with computational models.
Methodologically, the study combined behavioral testing with machine learning tools. The researchers used a pre-trained deep neural network to extract high-level auditory features from the recordings and then applied a support vector machine classifier to determine whether those features systematically differentiated hot and cold water. They also conducted psychophysical experiments to test whether listeners could consciously report temperature differences and whether they could implicitly discriminate temperature from sound.
The findings were consistent and robust: participants were able to distinguish hot from cold water by sound alone, often without conscious awareness that they were using thermal cues. This pattern suggests the perceptual ability is implicit—formed by repeated exposure to subtle physical differences in how water sounds at different temperatures rather than by explicit instruction.
Dr. Adi Snir, a postdoctoral fellow at the BCT Institute and co-author of the paper, emphasized that temperature perception differs from vision or hearing because it typically depends on specialized receptors in the skin. Yet, sensory systems can form cross-modal correspondences, and the present results show that audition can carry reliable information about thermal properties.
Prof. Amir Amedi, founding director of the BCT Institute, noted that previous behavioral reports hinted at this ability but left open the question of how and why it occurs. By pairing human behavioral data with machine learning classification, the team clarified which acoustic features correlate with temperature and demonstrated that a computational model can match human performance in classifying thermal properties from sound.
Beyond documenting the perceptual phenomenon, the study points to broader implications. Understanding auditory markers of temperature could improve designs for assistive devices, sensory substitution systems, or human–machine interfaces that leverage sound to convey environmental properties. The authors also propose future research to determine whether novel neural representations or sensory maps arise in the brain for these cross-modal associations, similar to established maps for vision and touch.
When asked about speculative extensions, Amedi mentioned that combining these methods with neural stimulation could, in principle, be used to augment or alter sensory experience—an idea that has been discussed in the context of neural interface technologies.
About this auditory neuroscience and AI research news
Author: Lital Ben Ari
Source: Reichman University
Contact: Lital Ben Ari – Reichman University
Image: The image is credited to Neuroscience News
Original Research: Open access. “Hearing temperatures: employing machine learning for elucidating the cross-modal perception of thermal properties through audition” by Adi Snir et al., published in Frontiers in Psychology.
Abstract
Hearing temperatures: employing machine learning for elucidating the cross-modal perception of thermal properties through audition
People can use auditory information to infer thermal properties, even when they report being unaware of this ability. Although most individuals deny that they can perceive the temperature of poured water by sound alone, behavioral testing and machine learning analysis show reliable discrimination of hot and cold water from the acoustic properties of pouring.
This cross-modal capacity appears to be implicitly learned over a lifetime of exposure to the physical differences in pouring dynamics at different temperatures. By combining psychophysical experiments with deep neural network feature extraction and classification, the study demonstrates that both humans and computational models can classify thermal properties from auditory cues and begins to characterize the underlying physical and perceptual features that enable this mapping.