Poornima Madhavan

Diversity during training enhances detection of novel stimuli

This research demonstrates that when individuals are expected to detect novel targets, they will be best prepared when trained with diverse categories. Participants were trained in a simulated luggage screening task in one of three conditions of diversity: high (participants searched for dangerous objects belonging to five different categories); low (participants searched for targets belonging to one of the five categories); and no training (control condition). After training, all participants were asked to look for the same novel dangerous objects in the bags. Results show that, during training, the low diversity condition resulted in highest hit rates and fastest response times. In contrast, after training, results were reversed: participants that trained in a high diversity condition were most effective at detecting novel targets. Those with no training at all were equally poor at detecting novel targets as those that trained in a low diversity condition.

A note of caution regarding anthropomorphism in HCI agents

Universal usability is an important component of HCI, particularly as companies promote their products in increasingly global markets to users with diverse cultural backgrounds. Successful anthropomorphic agents must have appropriate computer etiquette and nonverbal communication patterns. Because there are differences in etiquette, tone, formality, and colloquialisms across different user populations, it is unlikely that a generic anthropomorphic agent would be universally appealing. Additionally, because anthropomorphic characters are depicted as capable of human reasoning and possessing human motivations, users may ascribe undue trust in these agents. Trust is a complex construct that exerts an important role in a users interactions with an interface or system. Feelings and perceptions about an anthropomorphic agent may impact the construction of a mental model about a system, which may lead to inappropriate calibrations of automation trust that is based on an emotional connection with the anthropomorphic agent rather than on actual system performance.

The relationship between stimulus-response mappings and the detection of novel stimuli in a simulated luggage screening task

Automaticity research suggests that consistent mapping leads to better skill acquisition than varied mapping. Other research demonstrates that in some situations varied mapping leads to better transfer. The effect of stimulus-response mappings in complex visual inspection tasks, wherein transfer stimuli might differ from training stimuli, has seldom been studied. Therefore, the effects of consistency vs variability of practice on learning and transfer in a simulated luggage screening task were compared. Consistent mapping led to faster and more accurate initial skill acquisition. However, during transfer, varied mapping led to higher levels of sensitivities and confidence and fewer deviations from optimal response criteria. Consistent mapping assists initial skill acquisition; however, in tasks where the physical identity of transfer stimuli might differ from those used in training, varied mapping leads to more efficient transfer. The results provide an important starting point for training individuals to achieve optimal transfer of learning in complex tasks. This research demonstrates that a combination of variables influence transfer of learning in real-world visual inspection tasks wherein transfer conditions may not be identical to training conditions. Acquisition of skills during training is facilitated by consistent stimulus-response mappings; however, in order to ensure optimal transfer of skills to situations involving novel stimuli, training should incorporate varied mapping of stimulus-response elements.

Advice Taking from Humans and Machines: An fMRI and Effective Connectivity Study.

With new technological advances, advice can come from different sources such as machines or humans, but how individuals respond to such advice and the neural correlates involved need to be better understood. We combined functional MRI and multivariate Granger causality analysis with an X-ray luggage-screening task to investigate the neural basis and corresponding effective connectivity involved with advice utilization from agents framed as experts. Participants were asked to accept or reject good or bad advice from a human or machine agent with low reliability (high false alarm rate). We showed that unreliable advice decreased performance overall and participants interacting with the human agent had a greater depreciation of advice utilization during bad advice compared to the machine agent. These differences in advice utilization can be perceivably due to reevaluation of expectations arising from association of dispositional credibility for each agent. We demonstrated that differences in advice utilization engaged brain regions that may be associated with evaluation of personal characteristics and traits (precuneus, posterior cingulate cortex, temporoparietal junction) and interoception (posterior insula). We found that the right posterior insula and left precuneus were the drivers of the advice utilization network that were reciprocally connected to each other and also projected to all other regions. Our behavioral and neuroimaging results have significant implications for society because of progressions in technology and increased interactions with machines.

Differential Base Rate Training Influences Detection of Novel Targets in a Complex Visual Inspection Task

We studied the effects that multiple levels of signal probability (known as base rate) have on the transfer of learning in an airline luggage screening task. Participants (n = 33) were presented with three base rates during the acquisition (training) phase: 100%, 50%, or 20%; at transfer, all participants detected novel targets at a base rate of 20%. Performance was measured by hit rates, false alarm rates, sensitivities, and detection times. Results revealed that participants receiving higher base rates during training obtained higher hit rates at transfer compared to participants encountering lower base rates. However, increasing the training base rate also increased the incidence of false alarms, leading to a low overall level of sensitivity during transfer. Relatively higher base rates had mixed effects on response times. These results have implications for improving training modules for individuals in complex visual inspection tasks.

The effect of distractor modality and processing code on human-automation interaction

We examined the effect of distractor characteristics (modality and processing code) on visual search performance and interaction with an automated decision aid. Multiple Resource Theory suggests that concurrent tasks that are processed similarly (e.g. two visual tasks) will cause greater interference than tasks that are not (e.g., a visual and auditory task). The impact of tasks that share processing and perceptual demands and their interaction with human-automation interaction is not established. In order to examine this, participants completed two blocks of a luggage screening simulation with or without the assistance of an automated aid. For one block, participants performed a concurrent distractor task drawn from one of four combinations of modality and processing code: auditory-verbal; auditory-spatial; visual-verbal; visual-spatial. We measured sensitivity, criterion setting, perceived workload, system trust, perceived system reliability, compliance, reliance, and confidence. Participants demonstrated highest sensitivity when performing with an auditory-spatial secondary task. Automation compliance was higher when the auditory-spatial distraction was present versus absent; however, system trust was highest in the auditory-verbal condition. Confidence (when disagreeing with the aid) was also highest when the distractor was auditory. This study indicates that some forms of auditory ‘distractors’ may actually help performance; these results further contribute to understanding how distractions influence performance when operators interact with automation and have implications for improved work environment and system design.

The Role of Affective Valence and Task Uncertainty in Human-Automation Interaction

Mood can greatly influence how humans experience and react to the world around them. Therefore, it seems likely that mood can also influence interaction with automated decision systems and that this influence could be exacerbated by characteristics of a given task. To test these ideas, participants completed a series of complex visual search tasks with different levels of uncertainty. Half of the participants received negative mood manipulations and half received positive mood manipulations. Also, half of the participants received the assistance of an automated decision aid and half did not. Participants were assessed for system trust, confidence, sensitivity, and criterion setting. Results revealed that participants in a positive mood were more susceptible to the recommendations of the decision aid and were more confident than those in a negative mood. Automation presence tended to reduce the impact of task uncertainty on the dependent variables. These findings have implications for system design and implementation decisions.

Framing and Context Effects in Visual Search Training

Framed incentive structures and context effects may have training implications for applied visual search tasks such as airline luggage screening. Participants were trained with various incentive structures that focused, or were sensitive to, various signal detection outcomes. Also, participants were trained with different context representations (weapons or produce search). Twenty-four hours later, participants performed a transfer session in which the incentive structure and target set was unknown. Incentive structures that focused on negative outcomes (misses) led to a response bias that was closer to optimal compared to structures that focused on positive outcomes (hits). Task context affected response bias but had mixed effects on sensitivity. Results of this study may better inform the design of training and automated support for airline luggage screening and similar applied visual search tasks.

Role of an Anthropomorphic Agent in Environmental Decision Making

Computer agents can significantly impact the manner in which people perceive information and make choices. This study examined the effectiveness of an anthropomorphized computer agent (a Polar Bear) versus a simple text box at two levels of reliability on participants’ receptiveness to information about climate change and subsequent environmentally friendly behavior. Participants (N=110) answered questions about climate change while receiving the assistance of one of four computer agents comprising either an anthropomorphic agent (Polar Bear) or text box at 90% or 60% reliability. Results revealed that dependence on the anthropomorphic Polar Bear agent did not significantly differ from dependence on information conveyed via the Text Box agent; however, accuracy on the environmental quiz was greater for participants assisted by the Polar Bear agent under 90% reliability but lower than those assisted by the Text Box under 60% reliability. This study demonstrates that in some contexts the decision aiding advantages of anthropomorphic characters are selective and may be dependent largely on the agent’s demonstrated reliability at helping users reach their end goal.

The Effects of Affect and Inspection Duration on Decision Time and Confidence

A variety of factors modify decision making behavior. The current study examines how affective state and inspection duration impact decision time and decision confidence in a simulated luggage screening paradigm. Participants (N=200), from each of three “affect” groups—primed for anger, fear, or sadness— and a control group were tasked with detecting weapon targets with inspection durations of either two seconds (high time pressured inspection) or six seconds (low time pressured inspection). Results revealed a main effect for inspection duration on decision time, such that participants with more highly time-pressured inspections had longer decision latencies after the luggage image timed out. There were also main effects for inspection duration and affective condition on decision confidence, such that participants in the low time pressure group had greater decision confidence and participants in the fear group had lower decision confidence than those in the control group.