Article Preview
Top1. Introduction
An alarm works in a very simple way: it attracts attention and asks for human intervention. From a cognitive point of view, alarm effectiveness in capturing attention is influenced by physical salience, “the degree to which a stimulus is likely to attract attention based on its low-level properties and independently of the mental state of the observer” (Awh, Belopolsky, & Theeuwes, 2012). It strictly depends on the sensory modality that is activated.
The most salient and widely used alarm modality is the auditory one (Edworthy & Hellier, 2006). It is used in many high-workload safety-critical environments, such as during surgery (Edworthy & Hellier, 2006). Over the last thirty years, their use has become even more widespread also in other contexts because of a greater attention to safety (Edworthy & Hellier, 2006; Shah et al., 2015; Sousa et al., 2016). Auditory alarms showed many criticalities: they could be confusing, obtrusive and uninformative (Edworthy, 1994; Schmidt & Baysinger, 1986; Meredith & Edworthy, 1995). Instead of being a mechanism for improving the safety of the worker, this kind of alarms often increases workload, and produces a hostile work environment due to high rates of false, annoying, and inopportune alarms (Xiao & Seagull, 1999).
Given the limitation of auditory alarms, different alarm modalities have been used. In particular the tactile modality that showed to be especially effective when cognitive load is split on several sensory channels, like in driving collision avoidance tasks (Chun et al., 2012; Meng & Spence, 2015). Tactile alarms seem to maintain the efficiency and omni-directionality of the auditory stimulus, granting an adequate level of salience even if they are not suitable for collective risk (Wickens & McCarley, 2007).
In general, it has been proven that the design of an alarm system should be customized to the specific condition of use, in order to improve the human-alarm interaction, analyzing the ways alarms are integrated into the working places in a systemic perspective (Amaldi et al., 2007, Haas & Van Erp, 2014; Van Erp, Toet, & Janssen, 2015). The goal of the present work was to provide an alarm system that was effective in the specific working environment of a toxicology lab of a major hospital in Torino, Italy, in the context of an IOT-based safety project (Antonini, Boella, Calafiore et al., 2016). In this environment, workers might be intoxicated by harmful reagents leaked from their containers, or might be over exposed to potentially toxic substances. Existing sensors detect data related to the level of toxicity of the environment, and go off once the threshold is overcame. Thus, in order to better know the context where to situate the alarms, the authors started with a participatory activity analysis to highlight risks and discomfort experienced by the operators in current working practices (Trizio, Occelli & Re, 2017). The analysis highlighted the specificities of the lab working conditions: (1) the workers might be doing critical activities that should not be interrupted if not strictly necessary; (2) they are potentially exposed to collective but also to personal risks, their performance often diverges from the expected procedures. Specifically, the non-compliance with organizational policies and procedures might cause personal risks that can be under-estimated even by the experienced operators. Moreover, the workers must have both hands free, therefore cannot manipulate a phone or other devices. Finally, the study evidenced a lack of a customized management of emergency situations, with no differentiation of the treatment of different typologies of risks, firstly personal or collective risks. As auditory stimuli are currently mainly used for all kinds of risks, it emerges the need of a different aid for managing alarms, more sophisticated and adapted to the specific conditions of the lab, capable to: