“…For the most part, high levels of proportional feature binding accuracy were observed in baseline conditions across the three experiments. This reveals that when 23 Running head: Working memory in spoken instructions the participant accessed each action ÔchunkÕ (see also Gilchrist, Cowan, & NavehBenjamin, 2009;Naveh-Benjamin, Cowan, Kilb, & Chen, 2007) by recalling the correct movement, they were highly likely to then complete this chunk by producing the appropriate object (in terms of both color and form), potentially indicating that partial loss of binding information in this paradigm was relatively rare. Movementobject chunks may therefore be encoded, retained (possibly within the episodic buffer), and accessed at retrieval in an all-or-none manner, at least when working memory resources are not directed elsewhere.…”