The question of ‘g‘  and its consequences for various hypotheses about intelligence have long been controversial (Heene 2008; Horn 2006b; Schonemann 1981; 2005). The question of whether particular models–Spearman’s g (Spearman 1904), Thomson’s sampling model (Thomson 1916), or multiple ability models (Thurstone 1934) –can be empirically distinguished from one another has long made the field of the psychometrics of intelligence halt at nearly a standstill. However, recent work in developmental / cognitive psychology models of intelligence have reinvigorated research in the area, perhaps finally breaking new ground (Conway & Kovacs 2013, 2015; Kievit & Simpson-Kent 2019; van der Maas et. al 2006; van der Maas et. al 2011).
A number of papers have offered empirical evidence for the mutualism model over the typical uniform static ‘general intelligence’/g model (Hofman et. al 2018; Kan et. al 2019; Kievit et. al 2019; Kievit et. al 2019; Ou et. al 2019; Peng & Kievit 2019). The model has wide explanatory power and can be used to solve outstanding questions in the debate in the literature over race differences, the ontology of intelligence, and the relationship between factors and heritability (Kan 2012). 
Kovacs and Conway have attempted to merge cognitive psychology and the psychometrics of intelligence with their ‘process overlap theory’–similar to sampling/bonds theory (Bartholomew et. al 2009, 2013)–which takes research from neuroscience, psychometrics, cognitive psychology and intelligence research (Conway & Kovacs 2018; Kovacs & Conway 2016a, 2016b, 2019).
Other papers offer watershed developmental models of intelligence (Cannon & Keller 2006; Kievit et. al 2016; Fuhrmann et. al 2019) using neurocognitive data. Other proposed models include network models (van der Maas et. al 2017), which also seem to receive empirical support (Schmank et. al 2019). There exist more developmental, and less psychometrics theories, such as Ackerman’s (Ackerman 1988, 1996; Ackerman & Kanfer 2004), Anderson’s (Anderson 1992), and the very recent Demetriou & Spanoudis’ (Demetriou & Spanoudis 2018).
Finally, empirical evidence has provided support for the necessity of a developmental mode of intelligence. For instance, ergodicity assumptions have been shown to be violated (Fisher et. al 2018; Hunter et. al 2014; Molenaar 2007, 2008, 2010; Samuelson et. al 2017; Schmiedek et. al 2019) for intelligence, meaning that intra- and inter-individual processes are distinct and important to separate. Moreover, the variability of individuals performance varies massively and accordingly to exquisitely fine factors (Weber et. al 2018).
 For the relationship between ‘g‘ and other cognitive abilities, see Matzke et. al (2010).
 Gignac (2014) and Gignac (2016a) profess to present empirical evidence against mutualistic models, but see van der Maas & Kan (2016), here and here. Gignac (2016b) responds by appealing to parsimony, but see here (Roche 2018).