This Ezra series profiles recently hired faculty members across Cornell’s colleges, schools and units. In their own ways, these researchers, scholars and teachers embody the university’s creative and collaborative vitality.

 

 

Wendy Ju: Facilitating human-robot interactions

The act of crossing a busy street may not seem all that exciting, but to Wendy Ju, the human interaction between pedestrian and driver brims with hidden cues that help inform the development of autonomous vehicles and other robotic technologies.

An assistant professor of information science at the Jacobs Technion-Cornell Institute at Cornell Tech, Ju parses the subtleties of human interaction to better understand how robotics can be designed and integrated, safely and seamlessly, into our human world.

“If machines don’t have ‘tells’ like we do, that hint at what they will do next, everything feels like a surprise, an unwelcome surprise,” Ju says. “I look at how machines can give off enough information to make [these] interactions smoother.”

With autonomous vehicles, design decisions are a matter of life and death, Ju says. For example, her recent project, “Ghost Driver,” explored how pedestrians read, react and respond to driverless vehicles. The findings were a revelation to the automotive industry, which had thought that signs or lights installed on driverless vehicles would help cue pedestrians in on the car’s intent to, say, stop at a crosswalk.

In fact, such visual signals wouldn’t have the intended effect, researchers found, since pedestrians instinctively pay more attention to the car’s motion first, honing in on the wheels and bumper.

“Lots of people are good at designing user interfaces, but objects that have behaviors of their own – that’s an area where I have design intuition,” Ju says.

Ju’s research expands beyond the human-robot interactions between pedestrians and autonomous vehicles to include human-to-human interactions inside vehicles. Among her current collaborations is a project with information science colleague and human-robot interaction scholar Malte Jung, who heads the Robots in Groups Lab.

The vibrancy of Cornell Tech, together with the interdisciplinary nature of the Department of Information Science, has made Cornell a perfect fit for Ju and her research, she says.

– Louis DiPietro

Denis Willett: Learning the language of underground organisms

Denis Willett, assistant professor of entomology, runs an unusual kind of school: His pupils spend most of their class time underground.

They learn via chemical cues and only retain information for 48 hours. But a crash course is all they need to succeed in the real world.

Called entomopathogenic nematodes, these millimeter-sized worms infect and kill the insects that eat the roots of important crops, such as corn, onions and turfgrass. In the nematode school, Willett teaches the worms how to find insect pests by following the chemical distress signals that plants release.

“By teaching the nematodes to respond to these cues, we can increase our efficacy and put them into new environments,” he says. “It’s an alternative to pesticides in many cases.”

Nematodes are just one type of underground organism that Willett and his team study in the Applied Chemical Ecology and Technology Lab at Cornell AgriTech, part of the College of Agriculture and Life Sciences. Their goal is to understand how plants, insects and microorganisms interact with their environments. With expertise that spans chemistry, computer science, entomology, engineering and plant physiology, Willett and his team develop applied solutions that benefit agriculture. 

Willett, who joined AgriTech in 2018, specializes in applied insect chemical ecology. He uses robot-assisted technology to collect the smells that organisms produce, which allows him to essentially eavesdrop on the chemical “conversations” happening below ground and make decisions to support growers.

“Cornell in general, but also the AgriTech station, is a pretty remarkable place to do research,” he says. “The people are phenomenal, and the ideas and conversations really make it an exciting place to work.”

One of the other key organisms his lab studies is called phytophthora – a devastating pathogen that was responsible for the Irish potato famine of the 1840s. It has a mobile life stage, called a zoospore, that in some species swims in search of new plants to infect. This single-cell organism has no brain, yet it can sense chemical compounds and make choices about whether to move toward or away from them. 

“It can travel really, really fast. We calculated that if it were a human, it could do the space of a couple of football fields in a second,” Willett says. “We’re really excited about deciphering how this single cell makes decisions and how it navigates the world.”

– Jana Wiegand

Nika Haghtalab: Using technology for societal good

How can we use technology to create a force for societal good? Nika Haghtalab, assistant professor of computer science, explores this question – in particular, how machine learning algorithms can actually encourage people to invest in better lives.

“Can we think about the technological assessment systems we are building as a way to encourage society to be healthier, individuals to become better citizens, students to become better learners?” she asks. “Our systems have a dual role and responsibility in both assessing and defining qualifications.”

Coming to Cornell after receiving her doctorate from Carnegie Mellon and nearly a year as a postdoctoral researcher at Microsoft Research, Haghtalab chose Ithaca and the Faculty of Computing and Information Science because it is a pioneer in understanding human interactions in computing.

“Cornell had a vision of information science and computer science faculty collaborating way before any other universities were putting their money and efforts into weaving these two fields together,” she says, noting that this encompasses her particular research interests.

Originally intrigued by how computer software could learn about people, Haghtalab was also drawn to machine learning by its interdisciplinary nature. “Machine learning draws from many fields – theory of computation, statistics, artificial intelligence,” she says.

Known as a leader in the field of “tech fairness” and ethics, Cornell CIS is a good fit for Haghtalab, whose postdoctoral research focused on fairness and diversity as they relate to recruiting candidates through machine learning.

“The right search criteria can be used to recruit more diverse candidates in college admission and hiring,” she says. “The criteria that are currently used in practice mirror society’s unconscious and historic biases. The system we propose uses machine learning to look in the space of all attributes and suggest alternative criteria that accept more diverse qualified candidates.

“This is cool because our system automatically suggests criteria for increasing diversity that in some cases lawmakers and consultants had to spend months looking at the data to figure out.”

“What I really value about Cornell,” she adds, “is how it advocates building a theory with practical applications in mind. I work on machine learning problems that are motivated by real life, but I build a theoretical foundation for understanding them with provable guarantees.”

– Leslie Morris

Shaun Nichols: Using the tools of psychology to explore philosophical problems

People everywhere and in every age have encountered fundamental questions: What is right? What is wrong? How do I know the difference? Am I free to make choices?

Philosophers consider such questions and, in the traditional modes, hand down answers based on analysis and theory. But a recent movement among academic philosophers turns directly to the source – people – to consider these puzzles. 

A leader in experimental philosophy, Shaun Nichols uses psychological techniques to consider philosophical problems. He joined the faculty in fall 2019 as a professor in the Sage School of Philosophy in the College of Arts and Sciences, adding to Cornell’s strength in moral psychology.

“I think philosophy is borne of common sense,” Nichols says. “There are a lot of tools you can use to understand philosophical thought.”

He’s recently used economic games, textual analysis of etiquette manuals and learning-theory studies to locate the sources of moral beliefs. 

In one study, he used an economic game to show that people will punish a cheater even if the cheater won’t know that he has been punished. In another, he compared historical etiquette standards to discover which kinds of etiquette rules persist across time. He found that while many rules of etiquette go in and out of fashion, violations associated with disgust – such as prohibitions on spitting – endure.

 “Just as emotions play a role in which norms of etiquette get preserved, emotions also plausibly play a role in which moral norms get preserved,” Nichols says.

His current work draws on learning theory to try to understand how people acquire philosophically significant concepts and distinctions. He is collaborating with Tamar Kushnir, associate professor in child development in the College of Human Ecology, to study how children determine whether a rule applies universally or only to a certain group of people.

“All kinds of behaviors are available to us,” Nichols says. “How do we figure out whether those things are right or wrong?”

– Kate Blackwood

This story originally appeared in the online-only spring 2020 issue of Ezra magazine.