Man Talks to Machine

Exploring the intersection between technology and psychology

Why Virtual Humans Make Better Listeners

A new study reveals why virtual therapists may be coming soon.

In my blog comparing computerized therapy and human therapists, I listed the following as one advantage of the computer-based approach:

"When using a computer people may feel free to express thoughts that they are too ashamed or embarrassed to share with another person. The user knows that a computer will not judge them."

Danil Nevsky/Shutterstock
Researchers from the University of Southern California's Institute for Creative Technologies have published a study investigating this phenomenon. They found that people are, in fact, more open and honest when they don't believe a human is monitoring their communications.

Find a Therapist

Search for a mental health professional near you.

I have noticed this effect when comparing patient behavior in a live clinical setting to their behavior when using online treatment programs (e.g., this social anxiety program). However, the USC trial backs up my anecdotal observations with empirical data.

I contacted lead investigator Gale Lucas to find out more. Here are her responses to my questions:

1. Why do some people share personal details more openly with a computer than with another person?

Research suggests that computer-administered assessments remove at least two factors that have been shown to undermine anonymity and honest disclosure. The first is the presence of another person, and the second is the belief that responses are being observed in the moment.

First, merely being in the same room with another person can reduce anonymity and thereby decrease disclosure. For example, research has shown that when participants interact with each other from different rooms through a computer (like Skype), they feel more anonymous and disclose more, compared to when participants are face-to-face. That is, just sitting in the same room as another human being can make people respond less openly. Even if the other person doesn’t say anything, their mere presence can affect your behavior.

Second, even if they are not in the same room, the belief that another person is—in the moment—observing their responses can discourage people from disclosing. We can see this even without computers: Previous research has shown that people respond more openly in face-to-face interviews if the interviewer remains unaware of their answers, such as when they put their answers on a piece of paper rather than saying them out loud to the interviewer. When the answers are unobserved, people are willing to disclose more personal information.

Our recently published research adds to this previous finding. Indeed, we provide the first empirical evidence that virtual humans can increase disclosure in a clinical interview context by making participants feel though their responses are not currently being observed. All participants in our study completed an interview with a virtual human. However, some participants were led to believe that a researcher in the other room was watching their responses, and others were told that the virtual human was completely automated. The belief they were interacting with just a computer led participants to disclose more personal information. Indeed, participants said things like, This is way better than talking to a person. I don't really feel comfortable talking about personal stuff to other people, and, A human being would be judgmental. I shared a lot of personal things, and it was because of that.

2. An application mentioned in your study is using a virtual human to gather information from a patient, which is then reviewed by a doctor at a later time. Do you think the observed effect will still hold if the patient knows that a human will look at their responses in the future?

Our research, along with other research, suggests that the effect does hold if the respondent knows that a human will be looking at their responses in the future. Specifically, in our study, participants who interacted with a virtual human knew that they were being videotaped, that researchers would view the recording of their session later, and agreed to this procedure before they began the interview. Even though they knew that someone would watch their interview later, participants who believed they was nobody observing them in the moment revealed more during their interview with the virtual human.

The aforementioned findings from interviews that keep respondent answers confidential from the interviewer also support the argument that the effect will hold even when responses will be viewed later. For example, researchers have allowed respondents to answer the interviewer’s question by writing on a separate sheet of paper that is then sealed in an envelope. Even though the person knows their answers will eventually be seen by somebody, believing that responses are not being seen in the moment leads people to open up more than when the interviewer can view their answers during the interview.

3. What is the main advantage that human therapists hold over computerized therapy systems? From a technological perspective, do you think this problem is insurmountable?

There may be many reasons to use human therapists over computerized systems. We do not see computerized systems as replacements for human therapists. We instead see technology as tools that can be used to augment the therapy process. Moreover, our ability as researchers and developers to make computerized therapy systems is hindered by limitations in our understanding of how human therapists succeed in their work with clients. As valuable as it is to empirically address the question of what makes a good therapist, there are aspects of the therapeutic process that are difficult, if not impossible, to measure with our current techniques. As measurement techniques improve, we may get a better understanding of the process, which in turn can open up further possibilities for more advanced computerized therapy systems. However, I believe that there might be aspects of the therapeutic process that are beyond what we could ever measure. This may be the most compelling argument to use human therapists over computerized systems.



Fjola Helgadottir, PhD, is a senior research clinician at the University of Oxford and co-founder of AI-Therapy, a developer of fully automated CBT-based treatment programs. Follow her on twitter @drfjola.

Fjola Helgadottir, Ph.D., is a Senior Research Clinician at the University of Oxford, a registered clinical psychologist, technology enthusiast, and co-founder of AI-Therapy.


Subscribe to Man Talks to Machine

Current Issue

Let It Go!

It can take a radical reboot to get past old hurts and injustices.