In all panels, shaded circles and ovals represent neurons or neural populations; arrows with triangular heads indicate excitatory connections, and with flat heads, inhibitory connections. O1 and O2 represent two options. (A) Schematic of a model of decision-making adapted from Roe et al., 2001, Usher and McClelland, 2001; Usher and McClelland, 2004, Churchland and Ditterich, 2012. (B) Model of decision-making and working memory in a tactile delayed match-to-sample task adapted from Machens et al., 2005. F1 is the first stimulus presented (tactile stimulation), F2 is the stimulus presented after a delay. F1+/F2+ (F1-/F2-) represent neurons that respond with increasing (decreasing) firing rates for increasing tactile stimulation frequency. The (+) and (-) represent pools of decision making neurons which mutually inhibit each other via feedback inhibition. The control signal helps route information from the F+/F- pools of neurons to the decision-making pools. The two states of the control signal are depicted here with two colors – orange and cyan. (C) C1 shows a 2-option version of the competitive selection framework proposed here (and found in several species – see subsection titled ‘Plausibility: Biological…’). C2 shows a circuit equivalent of the model in C1 (Mysore and Knudsen, 2012) which can be reduced to the classic mutual inhibition model. C3 shows the high-level, mutual inhibition model that summarizes all the models discussed in A,B and C; same as Figure 2C. Note that C2 implements a combination of feedback inhibition and donut-like inhibition (Mahajan and Mysore, 2020). (D) A recurrent network model of decision-making and time-integration (Amit and Brunel, 1997; Wang, 2002; Lo and Wang, 2006; Wong and Wang, 2006; Furman and Wang, 2008; Louie et al., 2011; Deco et al., 2013) (Adapted from Wong and Wang, 2006). Under special assumptions of linear input-output functions and gating variables, this model also reduces to the classic mutual inhibition model in C3. I: denotes population of inhibitory neurons; NS: denotes population of neurons nonselective for either input.