Jonas M. B. Haslbeck

How well do network models predict observations? On the importance of predictability in network models

Network models are an increasingly popular way to abstract complex psychological phenomena. While studying the structure of network models has led to many important insights, little attention has been paid to how well they predict observations. This is despite the fact that predictability is crucial for judging the practical relevance of edges: for instance in clinical practice, predictability of a symptom indicates whether an intervention on that symptom through the symptom network is promising. We close this methodological gap by introducing nodewise predictability, which quantifies how well a given node can be predicted by all other nodes it is connected to in the network. In addition, we provide fully reproducible code examples of how to compute and visualize nodewise predictability both for cross-sectional and time series data.

Temporal dynamics of number-space interaction in line bisection: Comment on Cleland and Bull (2015).

Several lines of evidence have converged on the idea that numerical and spatial cognition are linked (Dehaene, Bossini, & Giraux, 1993; Hubbard, Piazza, Pinel, & Dehaene, 2005; Wood & Fischer, 2008). Bisecting a line, flanked by taskirrelevant cues, to indicate the line midpoint is one way to assess number–space interaction. When using non-symbolic numerical cues as flankers, a systematic bias away from the midpoint has been reported. While this bisection bias initially has been attributed to a cognitive illusion of length caused by magnitude information encoded in the flankers (de Hevia, Girelli, & Vallar, 2006; de Hevia & Spelke, 2009; Stöttinger, Anderson, Danckert, Frühholz, &Wood, 2012), an alternative explanation has suggested a prominent influence of the visual properties of the flankers (Gebuis & Gevers, 2011; Gebuis & Reynvoet, 2012, 2013). Cleland and Bull (2015; henceforth C&B) have recently shown that subtended area and aggregate surface area of the non-symbolic flankers largely explain the bisection bias. In line with Gebuis and Gevers (2011), they observed a reversed bias when controlling for these visual properties—that is, participants bisected the line closer towards the smaller number covering a larger area. This influence of area was also observed when flankers contained the same number of dots. C&B therefore suggested visual cues as primary driver of their bisection bias. In addition to the supposed influence of visual properties on bisection, higher cognitive processes could also affect line bisection. C&B speculated that prioritizing and weighting of different cues may take place, which would imply competing mental representations. However, paper-and-pencil designs usually employed in line bisection studies do not allow investigating the dynamics of the underlying processes across time, which is necessary to investigate the presence of competing mental representations. In the following, we will thus discuss a computerized bisection task as a tool for exploring the temporal dynamics of number–space interactions. Using a computer mouse for line bisection ensured a highly standardized paradigm (see Supplemental Material) and added a novel measure of the bias reflected in movement trajectories. The rationale behind our approach was based on past mouse tracking studies on response behaviour: if there is a conflict between competing