Context matters: Behavioural observations alone do not reliably capture emotional reactivity in Icelandic horses

Behavioural responses are widely used to assess emotional reactivity in horses, yet their interpretation depends on the context in which they occur and may not fully capture individual variation. To address this gap, emotional reactivity and fearfulness were investigated in 48 Icelandic horses through a four-stage test battery consisting of an acclimatisation period, an Approach test (AT), a Bridge test (BT), and a Startle test (ST). Behavioural reactions were scored on ordinal scales and heart rate (HR) was recorded. Investigative behaviour, fear- and frustration-related behaviours, and differences related to sex and age were explored. Correlation analyses revealed that behavioural reactions were not consistent across tests. Reaction scores in the BT and ST showed a weak positive correlation (r_s = 0.34, P = 0.02), but both correlated negatively with AT scores (BT vs. AT: r_s = −0.30, P = 0.04, ST vs. AT: r_s = −0.38, P = 0.009). Within tests, reaction scores correlated with HR in the ST (r_s = 0.65, P ' 0.001), but only showed a tendency in the BT (r_s = 0.34, P = 0.07) and no correlation in the AT (P ' 0.1). Mares had higher HR and expressed more fear- and frustration-related behaviours than geldings. Age was negatively associated with investigative behaviours, pointing to curiosity in young horses. A principal component analysis yielded two components explaining 61% of the variation, reflecting baseline arousal (31%) and stimulus-evoked emotional reactivity (30%). Collectively, these findings indicate that emotional reactivity and fearfulness in Icelandic horses vary across individuals and are influenced by sex, age, and previous experiences. The inconsistent correlations highlight that a multivariate approach is needed to capture individual variation in emotional reactivity.