Neurodynamic Evidence Supports a Forced-Excursion Model of Decision-Making under Speed/Accuracy Instructions

Abstract

Evolutionary pressures suggest that choices should be optimized to maximize rewards, by appropriately trading speed for accuracy. This speed-accuracy tradeoff (SAT) is commonly explained by variation in just the baseline-to-boundary distance, i.e., the excursion, of accumulation-to-bound models of perceptual decision-making. However, neural evidence is not consistent with this explanation. A compelling account of speeded choice should explain both overt behavior and the full range of associated brain signatures. Here, we reconcile seemingly contradictory behavioral and neural findings. In two variants of the same experiment, we triangulated upon the neural underpinnings of the SAT in the human brain using both EEG and transcranial magnetic stimulation (TMS). We found that distinct neural signals, namely the event-related potential (ERP) centroparietal positivity (CPP) and a smoothed motor-evoked potential (MEP) signal, which have both previously been shown to relate to decision-related accumulation, revealed qualitatively similar average neurodynamic profiles with only subtle differences between SAT conditions. These signals were then modelled from behavior by either incorporating traditional boundary variation or utilizing a forced excursion. These model variants are mathematically equivalent, in terms of their behavioral predictions, hence providing identical fits to correct and erroneous reaction time distributions. However, the forced-excursion version instantiates SAT via a more global change in parameters and implied neural activity, a process conceptually akin to, but mathematically distinct from, urgency. This variant better captured both ERP and MEP neural profiles, suggesting that the SAT may be implemented via neural gain modulation, and reconciling standard modelling approaches with human neural data.

The Neurodynamic Decision Variable in Human Multi-alternative Perceptual Choice

“…We therefore hypothesised that if changes in corticospinal excitability are driven by accumulation-to-bound dynamics as encapsulated in the LCA model, MEP changes associated with each response would display typical characteristics of a decision variable. Specifically, we expected the MEPs' build-up rate to increase with evidence strength, and their amplitudes to reach a stereotyped level at the time of response (Hadar et al, 2016;Spieser, Kohl, Forster, Bestmann & Yarrow, 2018). Importantly, beyond these typical accumulation-to-bound dynamics, we also predicted a reduced amplitude (for potential responses) when participants prepared a four-choice compared to a two-choice decision, as suggested by previously observed lower baseline neural firing rates in non-human primates.…”

“…The resulting MEP signal reflects corticospinal excitability and has previously been associated with decision-related evidence accumulation (Hadar et al, 2016;Spieser et al, 2018). Typically, in humans, neural correlates of decision-making produce a single waveform per decision, reflecting either a difference or sum of the hypothetical underlying accumulators (e.g.…”

“…Additionally, as a positive control, we crossed a manipulation of difficulty with our manipulation of number of choices in a 2x2 design. Difficulty negatively affects accumulation rate (Mulder, van Maanen, & Forstmann, 2014;Ratcliff & Rouder, 1998;Teodorescu & Usher, 2013) and this effect has been demonstrated in MEP signals for two-choice tasks (Hadar et al, 2016;Spieser et al, 2018). To manipulate task difficulty, the percentage of pixels of the dominant colour varied between 33% (easy) and 30% (hard).…”

Temporal expectation hastens decision onset but does not affect evidence quality

“…Our study is the first to complement mathematical modelling with detailed EEG analyses regarding the effects of temporal expectation on decisional and nondecisional process. Such neural evidence is critical, given reported discrepancies between results from sequential-sampling models and neural signatures of decision formation (McGovern et al, 2018;Spieser et al, 2018). These additional analyses of the CPP allowed us to examine the effect of temporal expectation on the onset and rate of decision formation in the brain (Loughnane et al, 2016), and yielded further support for the sensory encoding account.…”

“…Previous attempts to discriminate between the sensory encoding account and evidence quality account employed such model-based analyses of behavioral data. However, recent studies on perceptual decision-making have revealed that conclusions based on prominent sequentialsampling models can be falsified by complementary analyses of neural signatures of decision formation (McGovern, Hayes, Kelly, & O'Connell, 2018;Spieser, Kohl, Forster, Bestmann, & Yarrow, 2018), suggesting that it is critical to corroborate insights from modelling with neural evidence. Specifically, non-invasive human EEG recordings have identified a domain-general build-to-threshold signal, the centroparietal positivity (CPP), that reflects decision formation via the gradual accumulation of sensory evidence (O'Connell, Shadlen, Wong-Lin, & Kelly, 2018).…”