Human click-based echolocation: Effects of blindness and age, and real-life implications in a 10-week training program

Abstract

Understanding the factors that determine if a person can successfully learn a novel sensory skill is essential for understanding how the brain adapts to change, and for providing rehabilitative support for people with sensory loss. We report a training study investigating the effects of blindness and age on the learning of a complex auditory skill: click-based echolocation. Blind and sighted participants of various ages (21-79 yrs; median blind: 45 yrs; median sighted: 26 yrs) trained in 20 sessions over the course of 10 weeks in various practical and virtual navigation tasks. Blind participants also took part in a 3-month follow up survey assessing the effects of the training on their daily life. We found that both sighted and blind people improved considerably on all measures, and in some cases performed comparatively to expert echolocators at the end of training. Somewhat surprisingly, sighted people performed better than those who were blind in some cases, although our analyses suggest that this might be better explained by the younger age (or superior binaural hearing) of the sighted group. Importantly, however, neither age nor blindness was a limiting factor in participants' rate of learning (i.e. their difference in performance from the first to the final session) or in their ability to apply their echolocation skills to novel, untrained tasks. Furthermore, in the follow up survey, all participants who were blind reported improved mobility, and 83% reported better independence and wellbeing. Overall, our results suggest that the ability to learn click-based echolocation is not strongly limited by age or level of vision. This has positive implications for the rehabilitation of people with vision loss or in the early stages of progressive vision loss.

Fig 1. An illustration of the general procedure for data collection for a single participant.

Note that tasks were done in different orders across participants and sessions, and this flowchart is just for illustration. In each of the 20 training sessions, each participant performed four separate tasks—size discrimination, orientation perception, virtual maze navigation, and real indoor/outdoor navigation. Please see the main text for descriptions of each of the tasks. For practical reasons, the natural navigation task was always done either in the beginning or the end of a session (in approximately equal parts). The other, lab-based, tasks were run in different orders across participants, with participants being able to choose the order if they wished. An additional session was used to acquire basic measurements of participants’ hearing (audiometry and DFM/DCI hearing tests)–please see the main text for details. Note that for some participants these were acquired before training began, rather than after (as indicated in the figure). For blind participants, they were contacted 3 months following training completion to complete a survey.

Fig 2. Illustration of the apparatus used for the size discrimination task.

The (larger) reference disk was used in every trial, and placed either on the top or the bottom bar (here: on the bottom). One of the five (smaller) comparison disks was placed on the remaining, free bar.

Fig 3. Illustration of the apparatus used for the orientation perception task.

The inset on the left illustrates the different orientations the rectangle could be presented at, i.e. vertical, right side up (45°), horizontal, or left side up (135°), always facing the participant.

Fig 4. Illustration of spatial arrangements used to construct virtual spaces (T-mazes, U-mazes, Z-mazes) for the virtual navigation task.

In all mazes, the box represents the starting area and the dashed black line symbolises the end point which sounded echo-acoustically different because it had been constructed from corrugated plastic sheets. For each position (i.e. intersection in each route) there were eight sound recordings (0°–315° in 45° steps). To navigate, participants used the computer keyboard (inset on the right-hand side). Each press of the ‘W’ key would move the participant one step forward in the virtual maze and the ‘S’ key would move them one step backwards, but still facing in the same direction. Each press of the ‘D’ key would rotate the participant 45° in clockwise direction and the ‘A’ key would rotate them 45° anti-clockwise.

Fig 5. Proportion of correct answers across sessions in the control training task.

Chance level for proportion correct is.5. Data from SCs and BCs are shown as black and white circles respectively, with each symbol representing the average and error bars representing the standard error of the mean across participants.

Fig 6. Performance in the size training task.

Top panel: Proportion of correct answers across sessions. Chance level for proportion correct is.5. Bottom panel: Distance at which participants performed the task across sessions. Data from SCs and BCs are shown as black and white circles, respectively with each symbol representing the average and error bars representing the standard error of the mean across participants. Data from experts (n = 3) who completed only a single session without training and positioned at 100 cm distance are shown as grey circles. For comparison, they have been plotted at session 20.

Fig 7. Performance in the orientation training task.

Top panel: Proportion of correct answers across sessions. Chance level for proportion correct is.25. Bottom panel: Distance at which participants performed the task across sessions. Data from SCs and BCs are shown as black and white circles, respectively with each symbol representing the average and error bars representing the standard error of the mean across participants. Data from experts (n = 5) who completed only a single session without training and positioned at 100 cm distance are shown as grey circles. For comparison, they have been plotted at session 20.

Fig 8. The mean time taken (seconds) to complete various mazes in sessions 1–20.

In session 15, unpredictable starting orientations were introduced, along with a 15 s timeout when a collision occurred. This is represented by the dashed black line. Data from SCs and BCs are shown as black and white circles, respectively with each symbol representing the average and error bars representing the standard error of the mean across participants. Data from experts (n = 4) who completed only a single session without training are shown as grey circles. For comparison, they have been plotted at session 14.

Fig 9. The mean number of collisions made in sessions 1–20.

In session 15, unpredictable starting orientations were introduced, along with a 15 s timeout when a collision occurred. This is represented by the dashed black line. Data from SCs and BCs are shown as black and white circles, respectively with each symbol representing the average and error bars representing the standard error of the mean across participants. Data from experts (n = 4) who completed only a single session without training are shown as grey circles. For comparison, they have been plotted at session 14.

Fig 10. The average proportion of successful maze completions in sessions 1–20.

In session 15, unpredictable starting orientations were introduced, along with a 15 s timeout when a collision occurred. This is represented by the dashed black line. Data from SCs and BCs are shown as black and white circles respectively, with each symbol representing the average and error bars representing the standard error of the mean across participants. Data from experts (n = 4) who completed only a single session without training are shown as grey circles. For comparison, they have been plotted at session 14.

Fig 11

(A) Mean time taken (seconds) to navigate ‘old’ and ‘new’ mazes. (B). Mean number of collisions made when navigating ‘old’ and ‘new’ mazes. (C). Proportion of ‘old’ and ‘new’ mazes successfully navigated. Error bars represent the standard error of the mean across participants.

Fig 12. Scatter plots of performance improvement in practical tasks plotted against age, split by participant group.

(A) Improvement in distance for the size task plotted against age. (B) Improvement in proportion correct for the size task plotted against age. (C) Improvement in distance for the orientation task plotted against age. (D) Improvement in proportion correct for the orientation task plotted against age. Data for BCs and SCs are in white and black symbols, respectively. Correlations are indicated in each plot. There is no relationship between age and/or blindness and performance improvement in practical tasks.

Fig 13. Scatterplots for improvement in performance in the virtual navigation task.

Any significant correlations are indicated in each plot with an asterisk. (A) Improvement in successes for the virtual navigation task plotted against age. (B) Improvement in time taken to complete mazes for the virtual navigation task plotted against age. (C) Improvement in time taken to complete mazes for the virtual navigation task plotted against binaural hearing differences. (D) Improvement in number of collisions for the virtual navigation task plotted against age. (E) Improvement in number of collisions for the virtual navigation task plotted against binaural hearing differences. (F) Improvement in number of collisions for the virtual navigation task plotted against performance in the DFM test at 500 Hz. Data for blind and sighted participants are in white and black symbols, respectively. Correlations are indicated in each plot, with significance indicated as * = p < .05; ** = p < .01.