Another way of looking at this is 'what technologies are likely to progress' and then working out what combinations of them are possible.
Radar stealth and sensors: Progress will be made in countering stealth through having greater networking of radar systems, as well as other improvements to radar processing and use of a greater variety of radar spectrums. However, radar stealth will still be useful for improving the easy of electronic countermeasures and reducing engagement ranges. As a result designs will progressively incorporate refinements to radar stealth technologies.
IR/Optical stealth and sensors: There are limits to IR/Optical stealth, including the reality that the plume of hot air from a jet engine will be visible under many atmospheric conditions (even if the aircraft wasn't). Sensors will continue to improve in this regard. However, their utility will still be limited in bad weather, and the development of DIRCM/DEWs should limit their utility for missile seekers.
Missile engines: Refinements of technology for air-breathing missiles should lead to increased kinematic performance for the missile weight. This will increase the ZNE.
Missile size trends: The need to deal with stealth and improved countermeasures may lead to an increase in seeker size (especially if compound seekers are used) which may push missiles larger. However, internal stowage of missiles may also increase the push for smaller missiles.
Aerodynamics: Improvements in production of light weight sensors will allow increasing redundancy and reliance on active stabilisation - allowing the overall stability margin to continue to decline (even further below what the FBW revolution allows). This will allow a refinement of L/D ratios and a move towards a higher degree of taillessness.
Further speculation: This may be limited to some extent by a need to retain supersonic maneuverability to defeat missiles, however the combination of radar stealth creating shorter engagement ranges and air-breathing missiles increasing their ZNEs eventually kinematic defeat of missiles may cease to be viable, with a transition to stealth, countermeasures and DIRCM/DEWs as the primary defenses.
Engines: Progress will continue to be made on variable bypass ratios. This will reduce the amount of fuel weight required for a given range while retaining the capacity for supersonic performance. Development of hypersonic engines and designs may take place, but their relative inefficiency will limit how widespread they become, with supersonic cruise and supersonic dash designs continuing to be common.
Further speculation: I'm less certain about the use of liquid hydrogen as a fuel for fighter designs, it has merits but would substantially increase the internal volume of the design, and other synthetic fuels may be more likely (although liquid hydrogen should become common for transport aircraft).
Size vs. numbers: A larger airframe can have a higher volume to surface area ratio. This allows sustaining higher speeds, having higher payloads, more powerful DIRCM/DEW capabilities, and longer ranged sensors. Size and maintenance costs or complexity are also not directly correlated (e.g. maintaining a large twin-engined fighter can cost a similar amount as a medium sized twin-engined fighter). This makes large platforms attractive and we may see several 40 ton designs, especially as maneuverability requirements become relaxed (see above). However, smaller nations may still need to field an adequate number of fighters and thus designs in the 20 ton range will remain.
Networking: Both manned and unmanned platforms (as well as non-airborne platforms) will be integrated into sensor networks to better counter stealth and to provide redundant mid-course guidance to missiles fired from long ranges behind the sensors platforms.
Further speculation: Due to risks of jamming and spoofing networks will be layered, with smaller local networks integrated with each other prior to being integrated with other networks by longer ranged datalinks. While all systems will be increasingly integrated, there will also be identifiable discrete 'units' engaged in 'aerial combined arms'.
As for what will appear in each generation - a lot depends on how quickly these various technologies mature with respect to each other.
There is a whole conversation about how air-defenses will co-evolve - but I think that is enough for now.
