Page 14 - It' about time: Studying the Encoding of Duration

General introduction Models of duration encoding researchers have proposed different models to describe how information about 1 duration can be derived from sensory input. Often these models are modular, proposing dedicated mechanisms that extract the duration of an event (Ivry, 1996; Ivry & Schlerf, 2008; Matell & Meck, 2004; Miall, 1989). The oldest and more widespread models assume a dedicated clock-like timing system in the brain (Gibbon, 1977; Treisman, 1963). According to these models, the brain contains a pacemaker-like unit which generates pulses at a steady rate. When timing an event, these pulses are collected by a so-called accumulator from the moment the event starts until it ends. After this, the accumulated pulses provide a code for the duration of the event that can be stored and used to guide behavior. Other dedicated models have proposed mechanisms such as coincidence detection in banks of oscillators (Ivry, 1996; Matell & Meck, 2004; van Rijn, Gu, & Meck, 2014) or differential patterns of activity in a set of delay lines (Desmond & Moore, 1988). Another line of theories is based on the idea that dedicated systems for duration encoding are not required (Buonomano & Laje, 2010; Karmarkar & Buonomano, 2007). Instead, these theories focus on the intrinsic temporal properties of the sensory signals being processed by the brain. According to the State Dependent Network (SDN) model, duration is encoded implicitly in the way that activity in a (neural) network changes over time. Even when presented with a stimulus that does not change, the response of the neural network responding to that stimulus will change over time. As such, these changes in neural responses provide an implicit signal for the passage of time while a stimulus is present. By learning the network states that are associated with specific stimulus durations, the duration of stimuli can be interpreted without a dedicated system that extracts information about time. In other words, the SDN model propose that temporal information is encoded in spatiotemporal patterns of activity without the need to explicitly encode or store information about duration. Yet, another type of intrinsic duration model relies on the endogenous neural oscillations naturally present in our brain (Herbst & Landau, 2016). Similar to the SDN model, the change in activity associated with neural oscillations provides an implicit signal for the passage of time that could be leveraged to track the duration of events. 13