To better understand the difficulties of listeners with hearing loss and the performance of hearing aids (HA), there is a need to develop outcome measures that can uncover differences (between listeners or between HA signal processing schemes) at the kind of moderately positive SNRs that are prevalent in everyday situations. Standard speech tests are often not useful at these realistic SNRs because accuracy is near ceiling. Moreover, standard speech tests do not capture differences in how much effort HA wearers have to exert, a crucial factor for living with hearing loss.

Outcome measures that integrate accuracy and effort, a concept referred to as “listening efficiency”, hold the promise of being able to discriminate between listening conditions even if they show almost equal intelligibility. Listening efficiency has previously been implemented as the ratio of accuracy to response times, and more recently as differential evidence accumulation rates estimated in a cognitive model of decision making that fits accuracy and response times jointly. In this study, we aimed to explore the concept of listening efficiency and its ability to reveal effects of HA signal processing and SNR on the performance of HA wearers, using the auditory lexical decision task (LDT). This is a well-established task in experimental psychology, and because it measures word recognition, it taps into many of the same cognitive processes as the speech tests more typically used in audiology. Participants are presented with a mixture of words and nonwords and are asked to respond as quickly and as accurately as possible whether each is a real word. Response times may be taken to reflect their underlying capabilities and the effort they exert.

Nineteen experienced HA wearers completed an LDT presented over a fixed 60-dB HINT noise at an SNR of either 5 or 10 dB, with the test HAs set to an omnidirectional or a directional program. Participants also completed a HINT test and questionnaires about listening effort. LDT response times and accuracy, as well as the ratio of accuracy to response times, were analyzed with linear mixed models. The SNR and HA-program manipulations only had small effects, and analyzing the ratios did not reveal larger effects than analyzing the accuracy or the response times alone. Additional work to derive listening efficiency from evidence accumulation rates, and to assess whether it leads to enhanced sensitivity to the SNR and HA program manipulations, is ongoing and will be presented at the conference, along with reflections on the merits of the task.